Ben Goertzel, known for his influential work in artificial general intelligence (AGI), believes that achieving human-level AI is likely just 2 to 3 years in the future. He emphasizes the significant implications and potential risks that come with this advancement, cautioning that once such technology emerges, it could fundamentally reshape society. Goertzel's insights underline the urgency of preparing for both the opportunities and challenges that a fully realized AGI might present. His perspective serves as a reminder of the importance of ongoing discussions within the tech community about the ethical and practical considerations of advanced AI systems.

Ben Goertzel, dubbed the 'Father of AGI,' claims human-level AI is 2-3 years from realization and issues a warning about the implications.

A dark research chamber illuminated by warm amber spotlights, cables snaking across the floor toward a towering translucent neural-network brain pulsing with orange synapse flashes. In the foreground, a lone scientist’s silhouette adjusts vintage control dials on a metal console, long shadows stretching toward the rising AI mind—an ominous sense of a new intelligence about to awaken.

‘Father of AGI’ Ben Goertzel says human-level AI 2-3 years away, warns ‘once you have a…’ - The Times of India

- Ben Goertzel predicts human-level AI could be achieved in the next 2-3 years.  
- He emphasizes the potential risks associated with developing advanced artificial intelligence.  
- Goertzel warns about the ethical implications and responsibilities of creating AGI technology.

âFather of AGIâ Ben Goertzel says human-level AI 2-3 years away, warns âonce you have aâ¦â

Recent insights from Ben Goertzel, a pioneer in artificial general intelligence (AGI), suggest that we are on the cusp of a major transformation in the workforce, with the potential of AI to take over most jobs possibly just two to three years away. Speaking with Forbes, Goertzel foresees a time when "the vast majority of human jobs become obsolete" due to the emergence of fully autonomous, human-level AI.

Goertzel emphasized that while significant change is imminent, the transition will not occur instantaneously. Once a breakthrough in AGI is achieved, a gradual rollout and adaptation phase is expected — a process similar to how AI technologies gained acceptance over time, particularly following the rise of generative AI in 2022.

He did, however, highlight certain professions that seem more resilient to AI influence, notably education. He expressed confidence that roles such as teachers will continue to thrive as the human element remains essential in learning environments.

With the evolution of AI transforming the job landscape, Goertzel indicated that professionals should focus on acquiring specific skills to secure their future relevance. He articulated that it’s challenging to predict which professions will be at risk first, as conventional expectations may be upended — for instance, traditional roles in law and graphic arts might be replaced before more hands-on jobs like plumbing or electrical work.

Goertzel identified three key skills that will enhance employability as AI becomes more integrated into work processes:

1. **Strong Human Relationship-Building**: As AI progresses, maintaining and nurturing human connections will be vital. Goertzel stressed the importance of developing emotional intelligence, effective communication across various contexts, and active listening skills. He affirmed that roles reliant on human interaction, particularly in teaching, will persist even in a landscape dominated by AGI.

2. **Ability to Pivot**: In a rapidly changing work environment, professionals must be flexible and adaptable. Goertzel advised that individuals should continually update their skills, whether through formal education or self-directed learning, to stay relevant. Employers today prioritize adaptability, prompting workers to engage with industry developments and reflect on their potential impacts on their careers.

3. **Be Comfortable With Yourself**: Goertzel urged individuals to embrace self-awareness and cultivate a deep understanding of themselves. This personal insight can enhance connections with others, allowing for more meaningful interactions in both personal and professional realms.

In summary, as we stand on the brink of a new era shaped by AI advancements, embracing these skills will be crucial to navigate the forthcoming changes and ensure a successful future in the evolving workplace.

Prepare for an AI-driven future with these 3 skills — AGI pioneer warns that ‘human jobs will be outdated’

In a dimly lit, moody auditorium, a lone, sage-like figure stands at a sleek black lectern bathed in warm amber light. Behind him, the shadowy silhouette of a towering humanoid robot looms against a dark wall. Floating in midair before him are three glowing icons—a paintbrush, a handshake, and a brain-shaped lightbulb—symbolizing the creative, social, and critical thinking skills to “AI-proof” your future. No blue tones, just deep shadows and candle-lit warmth.

AI-proof your future with these 3 skills — Father of AGI says ‘human jobs will become obsolete’ - Mint

- AGI is expected to replace most human jobs within 2-3 years, says Ben Goertzel.  
- Educators and teaching roles remain critical and less likely to be affected by AGI.  
- Key skills for future jobs include emotional intelligence, adaptability, and self-awareness.  

Artificial intelligence is only two or three years away from replacing most human jobs.

Ben Goertzel, the father of AGI (artificial general intelligence), told Forbes that there is a fully independent human-level AI, which he believes is only 2-3 years away, "the vast majority of human jobs become obsolete".

“Prompt engineering was the most important thing two years ago—but not now,” Goertzel said.

However, he noted that this won't happen overnight — once an “AGI breakthrough” occurs, there will be a transitional period before it fully rolls out and is adopted at scale, the AI pioneer and computer scientist said.

He added that it would be very similar to how AI became mainstream over the years — with Gen AI in 2022, it took a little while for organisations to fully adopt and implement it.

Also Read | AI pioneer Yann LeCun rejects Anthropic’s 50 percent job loss prediction

Which jobs are safe? Ben Goertzel said a few roles, like educators and teachers, will remain intact – unaffected by the AGI because a human touch remains critical.

AI-proof skills for a better professional future Ben Goertzel said that with the rise in AI-adoption, skills are changing rapidly — "No one can accurately predict which jobs will go first. Like, the fact that lawyers and graphic artists are going obsolete before plumbers or electricians or research mathematicians," he told Forbes.

But there are at least three skills, which Goertzel said the workers will need, ahead of the “transitionary period”, to enable them to be fully useful and remain relevant in the future of work:

1. Strong Human Relationship-Building Goertzel said that once super intelligence does come, assuming it comes out beneficially for our species, “what we’re left with is ourselves, our own minds and bodies and our, our friends and family and our human relationships.”

He reiterated that teaching occupations will outlast the AI era even at the advanced stage of AGI and ASI.

The ability to have access to a human connection or decision-maker is most critical in the AI era, he said. In that case, Goertzel said it is best to focus on building skills like:

Emotional intelligence

Rapport-building

Communication skills (at all levels)

Active listening 2. Ability To Pivot You need to be “able to pivot rapidly and tap dance really fast,” Goertzel told Forbes. Adaptability and agility are among the top skills required by employers today, according to Coursera and the World Economic Forum reports.

What this means:

Learning new skills every month

Invest in an upskilling course, or study for free, at least once a year

Stay abreast of industry news and critically reflect on how that might impact your work and organisation It is important to change and adjust to remain in touch with reality and be relevant, to be prepared for what’s in store for you next.

The father of AGI noted that people who work in roles that can be performed by a computer smarter than you, you must be willing and ready to switch over to a career where human touch and connection remain critical.

3. Be Comfortable With Yourself Goertzel suggested that humans need to get to the place where we enjoy ourselves for who we are, embrace the beauty of what’s around us, be fully present in the moment, and understand ourselves intimately so we can connect better with others – beyond work domains.

AI-proof your future with these 3 skills — Father of AGI says ‘human jobs will become obsolete’

Google DeepMind has unveiled a groundbreaking framework for assessing Artificial General Intelligence (AGI), moving away from conventional benchmarks towards a more intricate and dimensional methodology. This new approach evaluates AI systems based on ten cognitive dimensions that align closely with human capabilities, such as perception, reasoning, and social cognition, providing a nuanced assessment of their strengths and weaknesses. For instance, an AI may excel in problem-solving yet struggle with meta-cognition or social awareness. AI Grid highlights that this framework presents a more thorough and transparent means of evaluating AGI compared to simplistic, single-score metrics.

This innovative framework is structured around a detailed three-stage evaluation process, which encompasses cognitive assessments, comparisons with human performance benchmarks, and the visualization of results through radar charts that illustrate cognitive profiles. Researchers will find actionable insights in this methodology, which critiques existing benchmarks while addressing challenges like assessing creativity and response speed. Further enhancing its value, a $200,000 Kaggle hackathon has been initiated to develop improved AGI evaluation techniques through community collaboration.

Key Elements of the Multidimensional Framework

In summary, Google DeepMind has introduced a sophisticated multidimensional framework aimed at evaluating AGI, concentrating on ten critical cognitive dimensions akin to human intelligence, such as perception and reasoning.

The evaluation process is threefold: first, cognitive assessments are conducted on specific tasks; second, AI performance is benchmarked against human data; and lastly, results are displayed in the form of cognitive profiles using radar charts.

However, several challenges persist in AGI evaluation, particularly concerning metrics for response speed, behavioral tendencies, creativity, and differentiating inherent intelligence from reliance on tools.

The hackathon seeks to crowdsource innovative solutions focused on five cognitive domains: learning, meta-cognition, attention, executive functions, and social cognition. This initiative aims to standardize AGI assessments, enhance transparency, and establish a rigorous scientific basis for tracking AI progress, addressing the current absence of a universal AGI definition.

At the heart of the framework lies a comprehensive cognitive taxonomy assessing AI systems on ten fundamental dimensions, including:

- **Perception**: Interpreting and processing sensory data.
- **Generation**: Producing coherent outputs like text and images.
- **Attention**: Focusing on relevant information while filtering distractions.
- **Learning**: Acquiring and adapting knowledge.
- **Memory**: Retaining and recalling information.
- **Reasoning**: Drawing logical conclusions and solving problems.
- **Meta-cognition**: Awareness and regulation of one's cognitive processes.
- **Executive Functions**: Skills associated with planning and decision-making.
- **Problem-Solving**: Identifying solutions to complex challenges.
- **Social Cognition**: Understanding social interactions and human behavior.

This multidimensional method prioritizes capabilities over mere outcomes, resulting in a detailed cognitive profile for each AI system. By emphasizing areas of strength and identifying gaps, the framework replaces simplistic evaluations—allowing a more nuanced view of AGI’s complexities.

A Rigorous Three-Stage Evaluation Protocol

The framework's three-stage evaluation protocol is crafted for comprehensive and reliable assessments:

1. **Cognitive Assessment**: AI systems are evaluated through targeted private tasks aimed at specific cognitive abilities, ensuring data integrity and result accuracy.
   
2. **Human Baselines**: AI performance is compared with relevant human samples to establish clear benchmarks, framing AI capabilities within the context of human cognition.

3. **Cognitive Profiles**: The final results are represented through radar charts, offering a visually intuitive and thorough portrayal of AI proficiency across the ten cognitive dimensions.

This structured evaluation strategy not only reveals where AI systems excel but also uncovers areas needing improvement in relation to human cognition. The framework furnishes researchers with significant insights, aiding in the development and enhancement of AI technologies.

Navigating Challenges in AGI Evaluation

While this framework marks a pivotal advance in AGI assessments, it also recognizes substantial challenges that warrant further consideration:

- **Response Speed**: Currently, the framework does not measure the speed of AI responses, an important factor in practical implementations.
  
- **Behavioral Tendencies**: Elements like risk preferences and alignment with human values are not explicitly assessed, even though they play a crucial role in ethical AI development.

- **Creativity**: The subjective nature of creativity complicates its definition and evaluation within AI systems.

- **Tool Usage**: Distinguishing between an AI's inherent capabilities and its dependence on external tools during assessments presents an ongoing challenge.

These issues highlight the necessity for continuous refinement of AGI evaluation methods to keep pace with evolving AI technologies.

Encouraging Community Engagement through Collaboration

To catalyze the formulation of new evaluation tasks, Google DeepMind has launched a $200,000 Kaggle hackathon, inviting participants from the global AI community to propose innovative methods for assessing five key cognitive dimensions. This collaborative effort aims to gather diverse ideas and approaches for advancing AGI evaluation.

The framework also critiques existing benchmarks, such as ARC AGI 3, which illustrate the hurdles AI systems face with novel reasoning tasks. By addressing these limitations, the new methodology aspires to convert subjective claims regarding AGI progress into quantifiable, evidence-based evaluations, promoting a transparent approach to AGI research.

Shaping the Future of AGI Research

As leading AI institutions like OpenAI and Google navigate the complexities of defining AGI, this framework emerges at a pivotal juncture. The absence of a universally accepted AGI definition complicates the measurement and comparison of advancements across various systems. By presenting a standardized multidimensional evaluation method, the framework seeks to bridge existing gaps and encourage collaboration in AGI research.

Ultimately, this initiative could redefine how AI advancements are conceptualized, measured, and communicated. By highlighting the "jagged frontier" of AI progress, the framework underscores the essential role of rigorous and transparent evaluation in steering development toward AGI responsibly. It represents a key step toward establishing a shared language for discussing and assessing AGI, contributing to the overarching mission of advancing AI in a safe, ethical, and scientifically informed manner.

Why Google DeepMind Recently Dropped Single-Score AI Evaluations

Warm natural light floods a sleek AI lab. At center, a massive analog gauge labeled “Single-Score AI Test” lies shattered on a polished table, its broken glass catching the sunlight. From the crack bloom dozens of smaller dials and color-coded charts marked “Fairness,” “Robustness,” “Transparency,” each glowing softly. A lone adult researcher, backlit by tall windows, reaches toward one new dial. On the far wall hangs a prominent Google DeepMind logo in bold colors.

Why Google DeepMind Just Abandoned Single-Score AI Testing - Geeky Gadgets

- Google DeepMind's new AGI evaluation framework assesses AI across ten cognitive dimensions, ensuring transparency.  
- The three-stage process includes cognitive assessments, human comparisons, and visual representations via radar charts.  
- Ongoing challenges in AGI evaluation include measuring creativity, response speed, and distinguishing tool reliance.  

Google DeepMind has introduced a new framework for evaluating Artificial General Intelligence (AGI), shifting from traditional benchmarks to a multidimensional approach. This framework examines AI systems across ten cognitive dimensions, including perception, reasoning and social cognition, to create a detailed profile of their capabilities. For example, an AI might demonstrate strong problem-solving skills but show limitations in areas like meta-cognition or social understanding. According to AI Grid, this method offers a more comprehensive and transparent way to assess AGI compared to single-score evaluations.

Explore the framework’s structured three-stage process, which includes cognitive assessments, comparisons to human baselines and the use of radar charts to visualize cognitive profiles. Learn about its emphasis on actionable insights for researchers, its critique of existing benchmarks and the challenges it acknowledges, such as measuring creativity or response speed. This guide also examines collaborative initiatives, including a $200,000 Kaggle hackathon, designed to refine AGI evaluation practices.

A Comprehensive Multidimensional Framework

TL;DR Key Takeaways : Google DeepMind has introduced a multidimensional framework to evaluate Artificial General Intelligence (AGI), focusing on ten cognitive dimensions that mirror human abilities, such as perception, reasoning and social cognition.

The framework employs a three-stage evaluation process: cognitive assessments on targeted tasks, benchmarking against human performance and visualizing results through cognitive profiles using radar charts.

Key challenges in AGI evaluation remain unresolved, including response speed, behavioral tendencies, creativity and distinguishing inherent intelligence from tool reliance.

A $200,000 Kaggle hackathon has been launched to crowdsource innovative solutions for evaluating five cognitive dimensions, fostering collaboration within the global AI community.

This initiative aims to standardize AGI evaluation, promote transparency and provide a science-based approach to measuring AI progress, addressing the lack of a universal AGI definition.

Central to this framework is a cognitive taxonomy that evaluates AI systems across ten critical dimensions, each representing a key aspect of human cognition:

Perception: The ability to interpret and process sensory information.

The ability to interpret and process sensory information. Generation: The capacity to create coherent outputs, such as text, images, or other forms of data.

The capacity to create coherent outputs, such as text, images, or other forms of data. Attention: The skill of focusing on relevant information while filtering out distractions.

The skill of focusing on relevant information while filtering out distractions. Learning: The ability to acquire and adapt knowledge over time.

The ability to acquire and adapt knowledge over time. Memory: The retention and recall of information for future use.

The retention and recall of information for future use. Reasoning: The capacity to draw logical conclusions and solve problems.

The capacity to draw logical conclusions and solve problems. Meta-cognition: The awareness and regulation of one’s own cognitive processes.

The awareness and regulation of one’s own cognitive processes. Executive Functions: Skills related to planning, decision-making and goal-oriented behavior.

Skills related to planning, decision-making and goal-oriented behavior. Problem-Solving: The ability to identify solutions to complex challenges.

The ability to identify solutions to complex challenges. Social Cognition: The understanding of social interactions and human behavior.

This multidimensional approach shifts the focus from how AI systems achieve results to what they are capable of accomplishing. By analyzing these dimensions, the framework generates a detailed cognitive profile for each AI system, highlighting areas of strength and identifying weaknesses. For example, an AI might demonstrate exceptional reasoning and memory capabilities but struggle with social cognition or meta-cognition. This method moves beyond simplistic, single-score evaluations, offering a richer and more accurate representation of AGI’s complexity.

A Structured Three-Stage Evaluation Process

To ensure thorough and reliable assessments, the framework employs a rigorous three-stage evaluation protocol. This structured process is designed to provide transparency and actionable insights into AI performance:

Cognitive Assessment: AI systems are tested on private, targeted tasks specifically designed to evaluate individual cognitive abilities. This approach minimizes the risk of data contamination and ensures the reliability of results.

AI systems are tested on private, targeted tasks specifically designed to evaluate individual cognitive abilities. This approach minimizes the risk of data contamination and ensures the reliability of results. Human Baselines: AI performance is directly compared to representative human samples, establishing a clear benchmark for measuring progress toward AGI. This comparison ensures that AI capabilities are evaluated in the context of human cognition.

AI performance is directly compared to representative human samples, establishing a clear benchmark for measuring progress toward AGI. This comparison ensures that AI capabilities are evaluated in the context of human cognition. Cognitive Profiles: The results are visualized using radar charts, offering an intuitive and comprehensive representation of AI performance across the ten cognitive dimensions.

This evaluation process not only highlights areas where AI systems excel but also identifies gaps where they fall short compared to human cognition. By providing a detailed analysis, the framework offers valuable insights for researchers and developers aiming to refine and improve AI systems.

Find more information on Google DeepMind by browsing our extensive range of articles, guides and tutorials.

Addressing Challenges in AGI Evaluation

While the framework represents a significant advancement in AGI evaluation, it also acknowledges several unresolved challenges that require further exploration:

Response Speed: The framework does not currently account for the speed at which AI systems generate responses, a critical factor in real-world applications.

The framework does not currently account for the speed at which AI systems generate responses, a critical factor in real-world applications. Behavioral Tendencies: Factors such as risk aversion and alignment with human values are not explicitly measured, despite their importance for safe and ethical AI deployment.

Factors such as risk aversion and alignment with human values are not explicitly measured, despite their importance for safe and ethical AI deployment. Creativity: Defining and evaluating creativity in AI remains an open question, as creativity is inherently subjective and context-dependent.

Defining and evaluating creativity in AI remains an open question, as creativity is inherently subjective and context-dependent. Tool Usage: Differentiating between an AI model’s inherent intelligence and its reliance on external tools during testing poses a significant challenge.

These limitations underscore the need for ongoing refinement of AGI evaluation methods to ensure they remain robust, relevant and adaptable as AI technologies continue to evolve.

Fostering Innovation Through Community Collaboration

To accelerate the development of new evaluation tasks, Google DeepMind has launched a $200,000 Kaggle hackathon. This initiative invites the global AI community to contribute innovative solutions for assessing five key cognitive dimensions: learning, meta-cognition, attention, executive functions and social cognition. By engaging a diverse range of participants, the hackathon aims to crowdsource creative and effective approaches to AGI evaluation.

The framework also critiques existing benchmarks, such as ARC AGI 3, which highlight the challenges AI systems face in novel reasoning tasks. By addressing these gaps, the new framework seeks to transform subjective claims about AGI progress into measurable, science-based assessments. This shift toward evidence-based evaluation is essential for advancing AGI research in a transparent and accountable manner.

Shaping the Future of AGI Research

This framework emerges at a critical moment in the development of AGI, as leading AI labs, including OpenAI, Google and Anthropic, continue to debate what constitutes AGI. The lack of a universal definition complicates efforts to measure and compare progress across different systems. By offering a standardized and multidimensional evaluation method, the framework aims to bridge this gap and foster greater transparency and collaboration in AGI research.

Looking ahead, this initiative has the potential to reshape how AI capabilities are understood, measured and communicated. By providing a clearer and more detailed picture of the “jagged frontier” of AI development, the framework emphasizes the importance of rigorous and transparent evaluation in guiding progress toward AGI responsibly. Ultimately, it represents a significant step toward establishing a common language for discussing and measuring AGI, contributing to the broader goal of advancing AI in a safe, ethical and scientifically grounded manner.

Media Credit: TheAIGRID



Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, Geeky Gadgets may earn an affiliate commission. Learn about our Disclosure Policy.

Why Google DeepMind Just Abandoned Single-Score AI Testing

Leading minds in artificial intelligence suggest that we are on the brink of a significant breakthrough. Dario Amodei, CEO of Anthropic, posits that within one to two years, AI could surpass human capabilities across various tasks. In his essay "The Adolescence of Technology," he warns of the emergence of Artificial General Intelligence (AGI)—an AI that can perform at human-level across a spectrum of domains.

Since the introduction of generative AI tools like ChatGPT in 2022, advancements in artificial intelligence have accelerated. Experts, including Amodei, predict we are nearing the era of AGI, where machines may outthink humans.

So, what exactly is AGI?

OpenAI describes AGI as AI systems exhibiting intelligence surpassing that of humans. In clearer terms, according to David Scott Krueger of Evitable, AGI represents an AI's capacity to perform any task a human can do. There is a consensus that AGI will signify a break from current AI technologies as we know them.

Current AI models, particularly large language models (LLMs), learn from extensive text databases to produce human-like text. Generative AI encompasses such models, which create new content—be it text, visuals, or programming code—based on data patterns. In contrast, AGI would entail a system capable of learning and applying knowledge in various fields similarly to humans.

How is AGI different from existing AI systems?

Melanie Mitchell, a computer scientist at the Santa Fe Institute, notes that the latest generative AI advancements have reignited discussions about the imminent arrival of AGI. However, despite their capabilities, these systems have not yet achieved AGI status. They lack the kind of fluid, adaptable reasoning that characterizes human thought. As Krueger explains, they struggle to maintain coherence over extended periods and can falter in real-world scenarios.

For instance, Tesla's self-driving software encountered issues where it inexplicably braked in certain locations due to misleading images, such as a billboard showing a police officer with a stop sign. Unlike humans, who can understand that a billboard is not a real stop signal, the AI did not grasp the distinction.

The pressing question is: “How well do these systems generalize beyond mere memorization?” suggests Raphaël Millière, a philosopher and cognitive scientist at the University of Oxford.

What advancements will AI need to achieve AGI?

Millière emphasizes that this is the "trillion-dollar question." One promising avenue is continual learning, enabling AI systems to evolve by learning from new experiences without discarding previous knowledge—much as humans do. For instance, ChatGPT does not refresh its core model based on user interactions.

Another crucial factor is data efficiency. Current AI systems often require vast datasets to understand concepts that humans can grasp with minimal examples. Millière highlights a demonstration where an image-generating AI could depict a pelican riding a bicycle but faltered when asked to reverse that scenario. Conversely, a young child could intuitively portray a bicycle riding a pelican without prior exposure. Thus, developing algorithms that can learn more efficiently from limited data may be essential for achieving true general intelligence.

There remain unresolved concerns regarding the path to AGI, including the ambiguity surrounding the necessary improvements in AI systems and the elusive nature of the finish line in this race.

How will researchers determine when AGI is achieved?

Industry leaders are actively pursuing answers. Companies like Meta, Anthropic, and OpenAI are dedicated to developing AGI, with different assessments of how success should be measured. Mitchell notes that while several tests have been proposed, none have proven satisfactory. Some experts suggest that a machine capable of navigating the internet to earn a million dollars could signify AGI, a definition Mitchell believes misses the essence of human-level intelligence.

The challenge in predicting AGI development stems from the difficulty in defining intelligence itself. Without a universally accepted standard to measure human-level intelligence, it becomes complicated to determine when AI reaches this threshold.

While Amodei anticipates AGI in a few years, there are skeptics who argue it could take much longer—or may never happen at all. Millière reflects on the unpredictable nature of AI advancements, citing a history of overly confident predictions that later proved inaccurate.

Moreover, the absence of consensus on the definition of AGI means it remains a shifting goal. In engineering terms, researchers are divided on the requirements for AI to match the versatility and efficiency of human cognition.

Why is the pursuit of AGI significant?

In theory, AGI could enable machines to perform tasks that current AI struggles with. For example, a true AGI system wouldn’t misinterpret a billboard as a stop sign. Mitchell illustrates this by comparing it to a robot designed to load a dishwasher. Although such robots exist, they struggle with unexpected scenarios—like a dog approaching and licking the dishes. An AGI capable of recognizing such changes and adapting accordingly would embody the flexible intelligence associated with human reasoning, rather than the rigid approach of today's AI systems.

AGI has the potential to revolutionize various aspects of life, with Millière noting that these systems could automate significant parts of the research process, thereby accelerating scientific discoveries—an application he sees as incredibly beneficial.

However, concerns regarding AGI persist. In 2023, the Future of Life Institute released an open letter urging AI laboratories to pause the development of powerful models beyond GPT-4. Influential figures like Yoshua Bengio, Elon Musk, and Steve Wozniak were among over 30,000 signatories.

Furthermore, the Center for AI Safety warned of the potential "risk of extinction from AI," also backed by notable figures in AI. This year saw Geoffrey Hinton depart from Google partly to voice concerns about losing control over AI systems that might surpass human ability.

What fuels this trepidation?

Key issues involve alignment—ensuring advanced systems adhere to human goals—and interpretability, as increasingly complex AI models can behave in unpredictable ways that challenge human understanding. Mitchell warns that overstating the capabilities of these systems may lead humans to relinquish decision-making authority, fostering a dangerous dependency.

The more autonomy we grant to intelligent machines, the less we retain ourselves. Krueger warns that once AGI is achieved, human control could diminish, potentially leading to profound disempowerment for humanity.

While opinions on the feasibility and timeline for AGI differ, the overarching theme remains uncertainty—an inherent element of the ongoing dialogue surrounding this pivotal technology.

Transitioning from AI to AGI

A bright, sunlit research lab with floor-to-ceiling windows overlooking a green courtyard. On a sleek glass table sit two brain sculptures side by side: the left is a metallic, circuit-patterned AI brain with tiny LED nodes; the right is a lifelike, organic-textured AGI brain with soft neural folds and copper filaments gradually merging into tissue. Between them, golden data particles drift as if caught in a gentle breeze. A focused adult scientist in a crisp lab coat stands to one side, jotting notes on a transparent clipboard. Behind, a whiteboard sketch reads “AI → AGI” with an arrow, bathed in warm natural light that fills the room with an atmosphere of hopeful innovation.

From AI to AGI - Scienceline

- Experts predict that AGI may be achieved within the next few years.  
- Current AI systems lack the adaptability and reasoning skills characteristic of AGI.  
- Concerns about AGI include potential loss of human control and misinterpretation of decisions.  

It’s only a matter of time, some of the smartest people in AI tell us. In as little as a year or two, artificial intelligence programs may outperform humans in a wide array of important tasks, Anthropic chief executive Dario Amodei recently declared. His prediction, made in his essay The Adolescence of Technology, was a warning about AI as capable as humans across many domains — which some experts would describe as Artificial General Intelligence, or AGI.

Artificial intelligence has advanced by leaps and bounds since generative AI tools like ChatGPT entered the market in 2022. Now, Amodei and others predict we will soon reach the point where machines can outthink humans: the dawn of AGI.

Here’s what you need to know.

What is AGI?

OpenAI defines it as AI systems that are generally smarter than humans. But Evitable CEO David Scott Krueger puts it this way: AGI is going to be an AI “able to do anything a human can.”

What they all agree on is that AGI would represent a fundamentally new kind of AI system.

Large language models (LLMs) are AI systems trained on vast amounts of text to generate human-like responses. They are a type of generative AI, which refers to systems that create new content — such as text, images or code — based on patterns in data. That includes systems like ChatGPT, Claude, or Gemini. AGI, by contrast, would be a form of AI able to understand and learn across many domains, adapting to new tasks much like a human can.

How does it differ from today’s AI systems?

The latest generative AI systems have “kind of revived” talk that we are closing in on AGI, says Melanie Mitchell, a computer scientist at the Santa Fe Institute. Even if these models seem very smart, they’re not quite considered AGI.

Despite their sophistication, these systems lack the flexible, autonomous and human-like reasoning that would define AGI. “They’re not able to remain as coherent as humans are over long time horizons,” says Krueger. “They’re not able to navigate the real world.”

Krueger says hallucinations — instances in which models generate information that isn’t true — remain a major limitation and are one reason these systems do not qualify as AGI.

Other current AI systems, such as those that are more task-oriented, provide a useful illustration.

Mitchell points to cases involving Tesla’s self-driving software, where drivers reported unexpected braking at the same location. “They noticed there was a billboard on the highway that had an advertisement with a police officer holding up a stop sign,” she says. Because the AI system had not encountered similar situations during training, it struggled to interpret the image correctly. “As humans, we know that a billboard is not a real stop,” says Mitchell, but the AI didn’t know.

The question is, “how much do they generalize, beyond memorization?” says Raphaël Millière, a philosopher and cognitive scientist at the University of Oxford.

What will AI systems need in order to achieve AGI?

“That’s the trillion-dollar question,” says Millière. Some people are betting on a concept called continual learning, he says. It refers to an AI system’s ability to keep learning from new data and experiences over time without forgetting what it previously learned — “like what humans and animals do,” he says. ChatGPT, for example, does not update its core model weights directly based on the user’s input.

“Another big one is data efficiency,” Millière says. Current models need an enormous amount of data to learn to do things that humans can learn from very few examples.

“For example, we asked an image generation model to draw a pelican riding a bicycle, and it could do that perfectly well. When we asked the LLM to draw a bicycle riding a pelican, however, it wasn’t able to do it. It kept generating a pelican riding a bicycle. When we asked a 7-year-old girl, she drew a very pretty bicycle riding a pelican,” explains Millière. She had never been asked to draw that before, but she could do it immediately. So, “the key to truly flexible general intelligence might also be better learning algorithms that can learn more efficiently from sparse data,” says Millière.

There is another problem, though. Not only is it unclear exactly how AI programs will need to improve in order to win the race to AGI, but no one is even sure where the finish line is.

How would researchers determine that AGI has been achieved?

Companies are working on it. Krueger says that Meta, Anthropic, OpenAI and others have been pretty clear that they’re working on developing AGI.

Experts have proposed different tests and benchmarks, but none of them are satisfactory, says Mitchell. “Some experts say that when a machine can go on the internet and figure out how to make a million dollars, that might be AGI. This seems to me like a misguided definition of human-level intelligence,” she says.

Part of the difficulty in predicting when AGI will be achieved is that even human intelligence is hard to define, and there is no single agreed-upon definition of it. Without a clear benchmark for what counts as “human-level” intelligence, it becomes difficult to determine when an AI system has truly reached it.

Amodei believes it’s a couple of years away, but not everyone thinks he’s right. Some skeptics argue it may take decades or even centuries to achieve AGI — if we ever get there.

According to Millière, predicting the trajectory of AI in general is difficult. “The historical record is littered with confident predictions that turned out to be mistaken,” he says.

There is also no agreed-upon definition of AGI, making it a moving target. And in terms of engineering, “Researchers fundamentally disagree on what will be needed for AI systems to match the flexibility, generality and efficiency of human intelligence,” Millière says.

Why would we want AGI, anyway?

AGI systems could, at least in principle, do useful things that current AI cannot. A true AGI system would not mistake a billboard for a real stop sign. Mitchell offers another example: a robot designed to load a dishwasher. “We have robots that already know how to do that,” she says.

But what if an unexpected situation arises? For instance, a dog wandering over and licking the dishes? A system able to recognize the change in context and decide to rinse the dishes again would illustrate the kind of flexible, adaptive intelligence associated with AGI, rather than the rigid task execution of conventional AI.

AGI could transform humans’ lives in significant ways. “I think AI systems are poised to automate significant aspects of the research process and to accelerate the pace of scientific discovery,” Millière says. In his opinion, this is one of the most promising candidates for a genuinely beneficial application of advanced AI.

Sounds good, so why are many experts so worried about AGI?

Concern about AGI is not new. In 2023, the Future of Life Institute published an open letter calling on AI labs to pause, for at least six months, the training of systems more powerful than GPT-4. Yoshua Bengio, Elon Musk and Steve Wozniak — three prominent figures in the field — were among more than 30,000 signatories.

The same year, the Center for AI Safety went further, stating the “risk of extinction from AI,” signed by Bengio, Sam Altman and Geoffrey Hinton. That same year, Hinton left Google, in part to speak more freely about these risks, warning of a loss of control and the possibility that AI could surpass human intelligence.

Why are they so worried?

Central issues include alignment — whether highly capable systems would reliably follow human intentions — as well as interpretability, since increasingly complex models may behave in ways that are difficult for humans to understand or predict.

One of the potential risks, Mitchell says, is that humans may overestimate the intelligence of these systems and will let them make decisions that they’re not really capable of making.

The more authority and autonomy we give to thinking machines, the less we will have. “It is clear that once we reach AGI, humans might no longer be in control, and we could face a disempowerment of humanity,” says Krueger.

Amodei has also warned that increasingly capable systems may be difficult to control. He writes in his essay that “humanity is about to be handed almost unimaginable power, and it is deeply unclear whether our social, political and technological systems possess the maturity to wield it.”

While views differ on AGI’s plausibility and timeline, uncertainty remains central to the debate.

From AI to AGI

Debate continues about the potential for artificial intelligence to reach human-level intelligence. However, one area where it has shown remarkable prowess is mathematics. Recently, A.I. systems developed by Google and OpenAI successfully tackled five out of six advanced problems at the International Math Olympiad, an esteemed competition that features the brightest high school mathematicians from around the globe.

Yet, despite its achievements, A.I. still exhibits considerable gaps in what many would consider basic common sense. Anuradha Weeraman, a software engineer from Sri Lanka, experienced this firsthand when he posed a seemingly simple question to various chatbots. He asked if he should walk or drive to a repair shop just 50 meters away, only to receive the advice to walk.

This peculiar phenomenon—where A.I. can excel in certain domains and flounder in others—is referred to as "jagged intelligence" by researchers, engineers, and economists. This term sheds light on A.I.’s uneven performance, highlighting its ability to excel in fields like mathematics and programming while struggling in more straightforward situations.

This concept is gaining traction among experts who develop and study A.I. systems, as it reframes the discussion about their intelligence. Rather than competing with human capabilities across the board, researchers contend that A.I. operates differently: it can outperform humans in specific tasks while underperforming in others.

What is ‘Jagged Intelligence’ and how can it reshape the conversation around AI?

A warm, sunlit studio scene: a classical stone bust of a human head lies fractured into jagged shards across a wooden pedestal. A sculptor’s hand hovers above, gently picking up one shard to reposition it. Soft morning light filters through a nearby window, casting sharp, angular shadows that emphasize the broken pieces and the emerging new form—capturing the idea of “Jagged Intelligence” and the promise of reframing how we see AI.

What Is ‘Jagged Intelligence’ and How Can It Reframe the AI Debate? - The New York Times

- A.I. demonstrated exceptional math skills at the International Math Olympiad, answering five complex questions.  
- A software engineer found A.I. struggled with simple common sense questions, like walking vs. driving.  
- "Jagged intelligence" describes A.I.'s varying performance across different tasks, excelling in some, failing in others.  

Say what you will about whether artificial intelligence will one day be as smart as a human. It has already become a star math student. Last summer, A.I. built by Google and OpenAI correctly answered five of six complex questions at the International Math Olympiad, an annual competition for the world’s top high school students.

A.I.’s common sense, however, may still be a bit lacking. A few months later, Anuradha Weeraman, a software engineer in Sri Lanka, noticed that leading A.I. systems struggled to answer what was essentially a trick question that most people would find laughably simple. When he told various chatbots that he needed to take his car to a repair shop that was only 50 meters away and asked if he should walk or drive, the bots told him to walk.

The strange way that A.I. looks like a genius at one moment and dense in another is what researchers, engineers and economists call “jagged intelligence.” They use this term to explain why A.I is racing ahead in some areas — like math and computer programming — while still struggling to make headway in others.

The term, which is widely used by the people building A.I. and analyzing its effects, could help reframe the debate over whether these systems are becoming as smart as, or even smarter than, humans. Instead, researchers argue, A.I. is something completely different: far better than humans at some tasks and far worse at others.

What Is ‘Jagged Intelligence’ and How Can It Reframe the AI Debate?

In a recent discussion on my podcast with Ben Goertzel, often referred to as "the father of AGI," he revealed some insightful projections about the future of work. Goertzel believes we could be just two or three years away from achieving true artificial general intelligence (AGI), which would rival or surpass human capabilities in the workplace. This sentiment echoes the thoughts of OpenAI's CEO, Sam Altman. 

Goertzel warns that with the advent of human-level AGI, many jobs may become obsolete. "Once you have AGI, the majority of human roles can vanish," he stated. However, he does not expect this shift to happen instantaneously. Instead, after an AGI breakthrough, there will be a period of transition, similar to how generative AI gradually integrated into various sectors since 2022.

He notes that certain positions, particularly in education, are likely to endure even as other roles disappear. Nonetheless, the timeline for significant workforce transformation is approaching rapidly, raising critical questions about how workers should prepare for potential job displacement.

As we approach this possible job disruption, it's essential for employees to start acquiring skills that will keep them relevant. Goertzel discussed how competencies can wane in importance and be replaced quickly. For example, what was once a coveted skill—like prompt engineering—has already lost its novelty.

When asked about the value of jobs beyond mere financial gain, Goertzel painted a vision of a utopian future where AGI handles most tasks. In this scenario, people might have more time for personal connections, family, and self-discovery, engaging in pursuits that provide emotional satisfaction rather than just financial reward. He suggests the idea of Universal Basic Income (UBI) and broader access to technology's benefits might become staples of this new society.

To make the most of this transition, Goertzel advises workers to focus on developing three key skills essential for thriving in an AI-dominant world:

1. **Human Relationship-Building and Communication**: As AGI progresses, what will truly matter is our humanity—our connections with others. Strong emotional intelligence, rapport-building, and effective communication will become vital, particularly in a landscape where machines handle most tasks. In an age of automatic resume filtering and AI job applications, the ability to forge personal connections and secure referrals is becoming increasingly important.

2. **Adaptability and Agility**: The ability to pivot quickly is crucial. As employers place a premium on flexibility, staying relevant means consistently learning new skills and remaining informed about industry developments. This might involve taking upskilling courses or transitioning to roles that prioritize human interaction—such as education or healthcare—over tasks performed by AI.

3. **Self-Comfort**: Surprisingly, Goertzel emphasized the importance of being comfortable with oneself, suggesting it’s vital to enjoy life beyond work. Embracing personal identity and appreciating the world around us can enhance our ability to connect meaningfully with others. This self-awareness is especially crucial as we navigate the uncertainty of the evolving job landscape.

As we contemplate the future of work, the focus may not solely be on which jobs will survive; instead, it is about the kind of individuals we will become in this new age.

The "Father of AGI" Advises Workers to Develop 3 Skills Immediately to Avoid Job Obsolescence.

A dark conference hall in deep shadows, a lone white-haired scientist in a suit standing on stage seen from behind under a single warm spotlight. In front of him hover three towering digital screens glowing softly: one with a stylized code bracket, one with a lit lightbulb, one with an intertwined heart-and-brain icon. In the audience’s silhouette, anxious office workers look up, emphasizing urgency as the “Father of AGI” warns them to build new skills before jobs disappear.

The “Father Of AGI” Warns Workers To Build 3 Skills Now Before Jobs Become Obsolete - Forbes

- AGI may render most human jobs obsolete within two to three years.  
- Workers should focus on emotional intelligence and adaptability to stay relevant.  
- Building strong human connections will become essential in a future dominated by AI.

When robots reach the point of being just as good as, or even better than, humans at work in a couple of years, the only thing that remains is human relationship-building and connection getty

What if the future of work, is…no work at all?

I had an eye-opening conversation with “the father of AGI,” Ben Goertzel, just yesterday on my podcast. He believes that we are a mere two or three years away from achieving fully independent human-level artificial intelligence, known as AGI (artificial general intelligence), similarly to OpenAI CEO Sam Altman.

And if it does, most jobs won’t survive, he warned. “Once you have a human-level AGI… the vast majority of human jobs become obsolete.”

Now, to be clear, the AI pioneer and computer scientist doesn’t believe this will happen overnight.

He expects that once an “AGI breakthrough” occurs, there will be a transitional period before it fully rolls out and is adopted at scale, pretty much how artificial intelligence went mainstream with Gen AI in 2022 but has taken a little while for organizations to fully adopt and implement.

He also believes that a few roles, like that of educators and teachers, will remain intact.

Nonetheless, a couple of years is not a long time at all for the start of a workplace upheaval, where robots can do most of the jobs of humans better than we can.

So in anticipation of a probable job apocalypse, what should workers be doing now?

MORE FOR YOU

Career Advice Is Swiftly Becoming Outdated

“Prompt engineering was the most important thing two years ago—but not now,” Goertzel said with a dry smile. Skills are quickly declining in value, he explained, giving way to other skills that rise in importance.

“Skills are changing rapidly, and no one can accurately predict which jobs will go first. Like, the fact that lawyers and graphic artists are going obsolete before plumbers or electricians or research mathematicians--I mean, that wasn’t obvious to anyone,” he remarked.

“But work is so much more than making money,” I questioned. “What about people who work just because they enjoy it and find it satisfying? What about those who work for the sense of pride, fulfillment, satisfaction and contribution to the economy?”

Goertzel then painted what he called a utopian world where:

AGI and superintelligence (next-level AGI) would be able to do most of our jobs

On the flip side of job loss, this would mean more time for human connection, spending time with loved ones and family, and getting to know ourselves better as humans

People could pursue more fulfilling and emotionally rewarding endeavors than chasing hours around the clock for money

UBI (Universal Basic Income) and distributed, decentralized access to the benefits of technology

When robots reach the point of being just as good as, or even better than, humans at work in a couple of years, the only thing that remains is human relationship-building and connection Rachel Wells

The 3 Skills You Need To Be AI-Proof

Ahead of the “transitionary period,” as Goetzel called it, workers will need to hone skills that will enable them to be fully useful and remain relevant in the future of work (if we can still call it that by that point):

1. Strong Human Relationship-Building And Communication Skills

“Once super intelligence does come, assuming it comes out beneficially for our species, you know, what we’re left with is ourselves, our own minds and bodies and our, our friends and family and our human relationships,” he said. “So that’s what matters. I mean, if you’re happy with yourself in your own life and you have people you love who love you, that will still be there and that’s probably the most important thing.”

Going back to my reference of teaching occupations as still outlasting the AI era even at the advanced stage of AGI and ASI, this actually makes a lot of sense.

The skills that will matter the most in the next frontier of AI will be the ones that are driven by human relationships.

Even today, we’re seeing this. In an era of AI slop resumes been churned out to recruiters, and ATS filtering out even the most capable candidates, the ability to have access to a human connection or decision-maker and get a referral is even more critical.

In that case, focus now on building skills like:

Emotional intelligence

Rapport-building

Communication skills (at all levels)

Active listening

Emotional intelligence and the ability to pivot quickly will save your career as we enter the new future of work getty

2. The Ability To Pivot

You need to be “able to pivot rapidly and tap dance really fast,” Goertzel told me. Adaptability and agility are in the top skills required by employers today, according to Coursera and the World Economic Forum reports.

If you’re used to sticking with the same way of doing things and refuse to adjust, even for a moment, you’ll be out of touch with reality and lose relevance, and won’t be prepared for what’s in store for you next.

This means:

Learning new skills every month

Invest in an upskilling course (or study free) at least once a year

Stay abreast of industry news and critically reflect on how this might impact your work and organization

And if you work in a role that can be performed by a computer smarter than you--let’s say accounting, for example--be willing and ready to switch over to a career where human touch and connection remain critical (like nurses, teachers, public speakers, certain types of coaching, etc.).

3. Be Comfortable With Yourself

This was an interesting take, as I didn’t expect this to be one of the skills Goertzel would mention. Nonetheless, I can see why it’s vital.

For example, I’m a self-proclaimed workaholic. And as such, “time for myself” often involves work to some degree.

I know others who would say that work is their identity, too.

But what Goertzel is suggesting here is that humans need to get to the place where we enjoy ourselves for who we are, embrace the beauty of what’s around us, and be fully present in the moment, and understand ourselves intimately so we can connect better with others--beyond work domains.

Quite a fascinating take on the future of work, for sure.

Because after all, it’s not a matter of, “What jobs will stick around?”

That part is still uncertain.

It’s more a matter of, “What’s the kind of person you will become in this new era?”

The “Father Of AGI” Warns Workers To Build 3 Skills Now Before Jobs Become Obsolete

In a seemingly mundane moment, I handed my daughter her meticulously polished CV, which I had carefully organized and refined from her original resume. After examining it for a moment longer than anticipated, she asked, “ChatGPT?” I affirmed her suspicion, and a wave of unexpected shame washed over me—not because I had misled her, but due to an instinctual feeling that I was crossing an unspoken boundary. I have always been forthright about my use of AI in my writing process, openly acknowledging its role in enhancing my work, even including this disclosure in a note to readers of my latest book.

This admission sparked a familiar tension—one that many writers seem to experience today. It reflects a modern form of vulnerability, akin to “coming out,” where initial guilt gives way to relief.

We find ourselves amidst a cultural shift. While AI has seamlessly woven itself into our daily fabric—drafting emails, correcting mistakes, summarizing discussions, and completing our sentences—it has simultaneously cast a shadow over the craft of writing. Now, a polished piece often raises eyebrows: if it seems too refined, too coherent, or too structured, it’s presumed to be the product of artificial intelligence. As a result, excellence leads to skepticism. Writers adapt not by honing their skills but by intentionally roughening their prose, removing stylistic flourishes, and opting for imperfections to convey authenticity.

This dynamic contributes to an atmosphere of quiet shame surrounding AI. While many people rely on it for drafting, brainstorming, and refining their work to meet pressing deadlines, they often respond to inquiries about their AI usage with self-defensive qualifiers: “Just a bit,” “Only for edits,” or “It’s mostly my own stuff.” Such qualifiers feel like apologies, emblematic of a broader reluctance to discuss AI use openly—not because they believe it’s morally wrong, but because they worry that others might judge them.

The underlying discomfort often isn’t about the technology itself but about identity. For generations, writing has been romanticized as a solitary endeavor where the struggle and effort signify authenticity. The advent of AI challenges this narrative: if technology can assist in crafting arguments and refining prose, it raises essential questions about ownership and authorship. The familiar markers of being a writer—the blank page, the rhythm of finding one’s voice, the struggle to express nuanced ideas—are now muddied by the incorporation of AI.

Critics have valid points, acknowledging that AI can diminish effort and potentially weaken cognitive skills. There are also significant concerns regarding originality and authorship. However, reducing the conversation to outright criticism overlooks the crucial distinction between augmentation and outright replacement. Utilizing AI as a collaborative tool does not equate to outsourcing thought. 

When I chose not to let ChatGPT create my daughter's CV from scratch, I was motivated by the desire for clarity and structure, based on my prior experiences with crafting CVs. The goal here wasn’t mere convenience; it was about enhancing an existing document, ensuring it was coherent and complete.

Public discourse on AI often polarizes between viewing it as a dire threat or a panacea. Yet, the reality for many lies in a middle ground: they utilize AI to enhance their productivity, creating clearer narratives while freeing themselves from tedious tasks without intent to deceive—merely to keep pace.

In professional contexts like medicine, where rapid analysis is vital, leveraging AI to sift through dense articles becomes indispensable for informed engagement. 

If the fear of AI prompts concealment, the issue lies not in the technology but in the surrounding culture. In environments where output is prioritized over processes, individuals may turn to “shadow AI,” using technology discreetly to evade scrutiny—not fostering open dialogue or collective progress.

This approach is unsustainable. It leads to isolation, hampers discussions, and impedes the development of robust standards for the proper use of AI. We need transparent expectations instead of judgments. Though some shame regarding AI usage is understandable, it often obscures vital conversations, stifling the dialogue we should be fostering.

Writing is transforming, not only altering how we create but also redefining authorship and compressing the relationship between effort and outcome. But applying the label of shame blurs critical discussions and stalls growth. Instead, what’s needed is openness and honesty about our interactions with AI.

So instead of asking, “Did you use AI?” a more productive question might be:

- “How did you engage with it?”
- “What unique contributions did you bring that the machine couldn’t?”
- “What aspects of your voice and judgment remain essential?”

These elements should still hold significance. The essence of writing—the choices regarding content and expression—should remain human-centered. The conversation should shift from whether we employ AI to a willingness to honestly explore how we use it and clarify what we can rightfully claim as our work.

Arthur Lazarus, previously a Doximity Fellow, is on the editorial board of the American Association for Physician Leadership and serves as an adjunct professor at the Lewis Katz School of Medicine at Temple University in Philadelphia. He has authored multiple books on narrative medicine and the Real Medicine fictional series, with his most recent work focusing on the intersection of artificial intelligence and medicine.

Artificial intelligence is transforming medical writing in the present day.

A sunlit modern office with a professional medical writer seated at a wooden desk, typing on a laptop. From the keyboard rises a glowing ribbon of clinical text that morphs into a DNA helix. Open medical journals, a stethoscope, and scattered prescription sketches lie nearby. Soft natural daylight streams through a nearby window, highlighting the fusion of AI-driven writing and scientific discovery.

Artificial intelligence is changing medical writing today - KevinMD.com

- Writers increasingly use AI for clarity, yet feel shame when admitting its use.  
- The perception of authenticity now emphasizes imperfection, complicating the writing process.  
- Conversations about AI's role in writing require openness rather than the stigma of secrecy.  

The moment was small, almost trivial. I handed my daughter her polished curriculum vitae (CV), clear, structured, and professionally phrased. All the information had been distilled and reorganized from her resume. She looked at it for a second longer than expected, then asked a single question: “ChatGPT?” I said yes, and felt an unexpected flicker of shame. Not because I had done something wrong. Not because I had tried to deceive her (I had not). In fact, I have been explicit in my writing that I use artificial intelligence (AI) to assist me with stylistic refinement, clarity, and phrasing of selected passages. I have said so publicly, even in a note to readers in my most recent book on artificial intelligence.

And yet, in that moment of admission to my daughter, I felt the same hesitation others describe in similar situations: a pause, a slight tightening, a sense that I had crossed some invisible line. That feeling is becoming increasingly common with writers I have spoken with, a modern-day equivalent of “coming out,” often accompanied by guilt, then relief.

The new social contract of writing

We are living through a strange cultural transition. On one hand, AI is everywhere, drafting emails, autocorrecting text messages, summarizing meetings, and completing sentences before we finish typing. It is deeply embedded in the infrastructure of personal and professional life. On the other hand, writing, particularly good writing, has become suspect. Too polished? Must be AI. Too structured? AI again. Too articulate? That uncanny voice of the machine. Too scripted? Definitely AI. Too many favorite words? A dead giveaway.

We have entered a peculiar era in which excellence invites suspicion. So, writers adapt, not by improving their craft, but by disguising it. They remove em dashes. They avoid overused words. They deliberately roughen their prose to signal humanness. Authenticity, once measured by a declarative voice, is now measured by imperfection.

The quiet shame of artificial intelligence

Employing AI results not only in suspicion but also silence. People use AI every day to draft, refine, brainstorm, and organize. They meet deadlines that would otherwise be impossible. They produce work that is sharper, clearer, and more coherent. But when asked, “Did you use AI?” the answer comes with qualifiers:

“Just a little.”

“Only for editing.”

“Mainly for research.”

“It’s mostly my own work.”

The qualifier functions as an apology. This dynamic has been described as a kind of “quiet shame,” where writers lower their voices when discussing AI, not because they believe it is wrong, but because they sense that others might. What is striking is not the use of the tool, but the reluctance to acknowledge it openly.

What are we really afraid of?

The discomfort is not really about AI in my opinion. It is about identity. For generations, writing has been tied to a particular mythology: the solitary thinker, wrestling with language, producing something uniquely their own. Difficulty was part of the credential. The struggle validated the outcome.

The use of AI is at odds with that narrative. If a machine can help structure an argument, refine a sentence, or suggest a more elegant phrasing, then what exactly are we claiming when we say we wrote this? This is the question that lingers beneath the surface. Not technological, but existential. Because AI removes the familiar markers of authorship: the emptiness of staring at a blank page, the idiosyncratic cadence of a sentence finding its shape, the struggle between thought and language that once felt like proof of ownership.

The critics and their valid concerns

It is true that AI reduces effort. That is, in part, the point. It is also true that overreliance may blunt certain cognitive skills. And there are legitimate concerns about originality, attribution, and intellectual rigor. But critique becomes distortion when it condenses everything into misuse, when using AI to refine a paragraph is equated with outsourcing thought, or when assistance is labeled as deception. That way of viewing AI misses a more important distinction: augmentation versus substitution. Using AI to think with is not the same as using AI to think for you.

Why not turn to ChatGPT to create my daughter’s CV? Perhaps it was the easier path. But I have done the exercise many times before, and there was little to be gained by repeating it. In this instance, the value lay not in generating the document from scratch, but in ensuring its clarity, structure, and completeness.

The middle ground most people occupy

The public discourse often splits into extremes:

AI as an existential threat

AI as an unqualified good

Most people, however, are not living at either extreme. They occupy a middle ground. They are curious but cautious. Productive but uneasy. Open-minded but unsettled. They are using AI not to replace their work, but to make it manageable with faster drafts, clearer narrative, and fewer hours lost to mechanical tasks. They are not trying to deceive anyone. They are trying to keep up.

In medicine, analytic speed is not a luxury. It is essential to navigating dense research articles, summarizing them, distilling key points, and determining their relevance before engaging more deeply in the topic.

A culture problem, not a technology problem

If people feel the need to hide their use of AI, the problem is not the tool. It is the culture surrounding it. In environments where output is valued but process is policed, people will inevitably resort to what some have called “shadow AI,” using the technology silently, without disclosure, without shared norms, and without collective learning.

This way of operating is not sustainable. It isolates individuals, prevents open discussion, and slows the development of thoughtful standards. What we need instead are explicit expectations and shared understanding, not judgment. Still, the shame may be partly justified, but not in the way we think.

There should be discomfort, because something important changes whenever we use AI. Writing is evolving. Authorship is being redefined. The relationship between effort and output is being compressed. But shame is a poor label. It stigmatizes more than it clarifies. It stifles conversation when what we need is precisely the opposite: openness and honesty about how we are working with AI.

A more honest framing of authorship

Perhaps the better question is not whether you used AI, but rather:

How did you use it?

What did you contribute that the machine could not?

What makes your prose meaningfully yours?

Where does your judgment, your experience, your voice still matter?

Because it still should count. The final act of writing, the act of deciding what stays, what goes, what is true, what is worth saying, should remain uniquely human. Otherwise, we risk claiming authorship without accountability. The question is no longer whether we use AI but whether we are willing to be honest about how we use it and what we can legitimately claim as our own.

Arthur Lazarus is a former Doximity Fellow, a member of the editorial board of the American Association for Physician Leadership, and an adjunct professor of psychiatry at the Lewis Katz School of Medicine at Temple University in Philadelphia. He is the author of several books on narrative medicine and the fictional series Real Medicine, Unreal Stories. His most recent book is Artificial Intelligence in Medicine: Controversy and Commentary.

Artificial intelligence is changing medical writing today

A growing number of investors assert that artificial general intelligence (AGI) has already arrived. If this holds true, we can anticipate significant transformations across various sectors, and one key area that may witness these changes initially is the Ethereum ecosystem.

On April 5, Marc Andreessen, co-founder of the influential venture capital firm Andreessen Horowitz, commonly known as a16z, took to social media to proclaim that "AGI is already here -- it's just not evenly distributed yet." AGI signifies AI technologies capable of performing tasks with human-level proficiency across diverse intellectual domains. Whether Andreessen’s assertion is premature or a forward-looking insight remains to be seen, but investors are keenly interested in where they might first experience the economic advantages of AGI.

There is speculation that cryptocurrencies, particularly Ethereum, could be ripe for AGI applications. Complicated KYC (know-your-customer) rules and compliance costs in traditional financial institutions create obstacles for automated systems. In contrast, an AI agent equipped with a crypto wallet can conduct transactions without needing explicit approval. Over the past year, Ethereum has been crafting its infrastructure in anticipation of this shift, which merits exploration.

The Ethereum Foundation, responsible for overseeing the network's development, established a specialized AI unit called dAI in September 2025, aimed at positioning Ethereum as the settlement layer for autonomous agents that can operate and collaborate independently of human oversight. This vision came to fruition in January 2026 when Ethereum launched a new standard for AI agents, known as ERC-8004, on its mainnet. This standard introduces three data registries to track identity, reputation, and work validation for agents, effectively establishing a trust framework for AI interactions with unfamiliar parties. As of April 8, over 37,000 agents had registered on the Ethereum platform, marking an essential step towards a functioning AI-driven economy.

Moreover, Ethereum enables these agents to engage in real-time micropayments without necessitating fresh accounts or wallets at each transaction, thanks to a protocol dubbed x402. The Ethereum ecosystem also boasts substantial on-chain capital: its decentralized finance (DeFi) sector holds around $54 billion in total value locked (TVL) alongside $165 billion in stablecoin liquidity.

With the combination of available capital, the x402 protocol, and the ERC-8004 standard, Ethereum currently possesses most of the essential elements for establishing an agent-based economy. In theory, this bolsters the intrinsic value of the cryptocurrency. However, the sophistication level of the agents remains critical as they do not yet exhibit the characteristics of AGI. Consequently, the primary demand for their capabilities is expected to originate from human users, limiting the ecosystem’s growth.

Looking ahead, if AGI is indeed on the horizon, the implications for Ethereum could be substantial. Agents equipped with genuine AGI functionality would likely require more financial resources compared to existing, simpler AI agents. By allowing more complex agents to earn and manage funds, their capability to procure services and undertake more complex tasks—independently—would likely increase. A significant portion of the economic activity generated by such powerful agents could take place on the Ethereum platform, leading to heightened interest and demand for the coin, particularly regarding transaction expenses.

If AGI is indeed a reality, Ethereum stands to benefit significantly, especially as a potential leading platform for commerce carried out by agents. However, a crucial element still appears absent: consistent demand that would facilitate agents engaging in economic activities at a scale needed to affect Ethereum's network traffic positively.

Therefore, while Ethereum's positioning in the AI landscape is intriguing and warrants close observation, it may not be an immediate buy signal. The emergence of an agent economy driven by AGI could unlock tremendous value, but if it fails to materialize, investors might avoid overpaying for a future that is further off than some experts predict.

When considering investing in Ethereum, it's wise to note that the Motley Fool Stock Advisor team recently identified ten other stocks viewed as better current investments than Ethereum itself. These selections have shown remarkable return potential in past years. Remember, historically, early adopters of stocks like Netflix and Nvidia, when recommended, saw transformative returns. Notably, Stock Advisor's average return significantly outpaces the S&P 500’s performance, emphasizing the importance of informed investment choices. 

Alex Carchidi holds positions in Ethereum, while the Motley Fool has financial interests in the cryptocurrency.

Marc Andreessen Claims Artificial General Intelligence (AGI) Has Arrived—Implications for Ethereum Explained.

A sunlit modern office where a figure reminiscent of Marc Andreessen, glasses perched and hands lightly clasped, gazes at a sculpted digital brain pulsing above a large metallic Ethereum logo on a sleek desk. Warm natural light floods through floor-to-ceiling windows, highlighting circuits in the brain and casting a sense of limitless AGI-driven innovation.

Marc Andreesen Just Said That Artificial General Intelligence (AGI) Is Here. Here's What That Could Mean for Ethereum. - AOL.com

- Marc Andreessen claims artificial general intelligence (AGI) is already here and needs distribution.  
- Ethereum's infrastructure supports autonomous AI agents in a potential agent-based economy.  
- The need for AGI-level capabilities could drive increased economic activity on Ethereum's platform.  

Key Points

Some investors believe that artificial general intelligence (AGI) is here.

If that's true, a lot is about to change.

Ethereum's ecosystem is one place where changes might start to occur first.

10 stocks we like better than Ethereum ›

On April 5, Marc Andreessen, co-founder of the famous venture capital firm Andreessen Horowitz, also known as a16z, declared on social media that "AGI is already here -- it's just not evenly distributed yet." AGI, or artificial general intelligence, refers to AI systems that are capable of matching human-level performance across a wide range of intellectual tasks. It's unclear whether his claim is premature or prophetic, but for investors, the main question is where an economic benefit from AGI would occur first.

There's an argument that cryptocurrency is one such place. Traditional financial venues impose know-your-customer (KYC) regulations and reporting requirements that create friction that autonomous software can't easily navigate, whereas an artificial intelligence agent with a crypto wallet can transact without permission. Ethereum (CRYPTO: ETH) spent the past year building infrastructure for exactly that scenario, so let's take a look at what the implications for it could be.

Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »

A pile of coins embossed with the Ethereum logo.

Image source: Getty Images.

Laying rails for a machine economy

The Ethereum Foundation, the nonprofit governing the network's development and its ecosystem, created a dedicated AI team called dAI in September 2025, seeking to make Ethereum the settlement layer for autonomous agents that pay and collaborate without any human intermediaries.

That ambition produced a concrete deliverable in January 2026, when the chain's new standard for AI agents, ERC-8004, went live on mainnet. ERC-8004 creates three data registries that track identity, reputation, and work validation criteria for agents, thereby creating a trust layer for AI that's transacting with unfamiliar counterparties. As of April 8, more than 37,000 agents were registered on Ethereum. So now Ethereum has a key part of the technical infrastructure needed to make an AI-based on-chain economy possible, and it also has a decent population of agents operating.

It also has the means to allow agents to actually exchange funds for services or goods. A protocol called x402 enables micropayments between agents in real time without the need to create any new accounts or wallets at the point of sale.

Furthermore, Ethereum brings a deep pool of on-chain capital for agents to tap. Its decentralized finance (DeFi) ecosystem holds $54 billion in total value locked (TVL), and $165 billion in stablecoin capital.

Between that base of capital, x402, and ERC-8004, the network thus has nearly all of the ingredients needed for a basic agent economy today. In principle, that makes the coin a bit more valuable than it would be otherwise.

The main missing ingredient is the sophistication of the agents themselves, which do not currently meet the criteria for being or having AGI. Without it, the main source of demand for their capabilities will come from human users, and the bulk of the goods and services that will be supplied to the market will be the result of a human's design and deployment, which limits the ecosystem's growth potential.

Advertisement

There could be plenty of growth on the way

Now, let's turn to the question of what artificial general intelligence would mean for Ethereum, specifically given its positioning, and assuming that Andreessen is correct.

Agents with actual AGI-level capabilities would likely have a larger need for financial resources than the less-complex AI agents that we've seen in operation so far. Equipping the stronger agents with funds, or allowing them to earn their own money, would be helpful because it'd enable them to purchase more services and take on larger and larger tasks with a higher level of independence than they'd be capable of otherwise. If a lot of the economic activity associated with those powerful agents occurs on Ethereum, it'd likely attract a lot more capital to the chain, and boost demand for the coin somewhat, as it's necessary to pay for transaction costs.

So if AGI really has been invented, there's likely some upside in store, and potentially a lot if Ethereum becomes the dominant platform for agent-based commerce. But again, for the moment, what's missing is sources of demand that would lead to actual agents doing economic work at a scale that moves the needle for Ethereum's network usage.

Thus, Ethereum's AI positioning is a thesis worth tracking rather than a buy signal today. If the agent economy materializes thanks to the development of AGI, it could capture enormous value. If it doesn't, you'll have avoided paying a premium for a future that was further away than what some of the smartest thinkers assumed.

Should you buy stock in Ethereum right now?

Before you buy stock in Ethereum, consider this:

The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Ethereum wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.

Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $556,335!* Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,160,572!*

Now, it’s worth noting Stock Advisor’s total average return is 975% — a market-crushing outperformance compared to 193% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.

See the 10 stocks »

*Stock Advisor returns as of April 14, 2026.

Alex Carchidi has positions in Ethereum. The Motley Fool has positions in and recommends Ethereum. The Motley Fool has a disclosure policy.

Marc Andreesen Just Said That Artificial General Intelligence (AGI) Is Here. Here's What That Could Mean for Ethereum.

On April 5, Marc Andreessen, co-founder of the renowned venture capital firm Andreessen Horowitz, commonly referred to as a16z, asserted via social media that "AGI is already here -- it's just not evenly distributed yet." AGI, or artificial general intelligence, pertains to AI systems that can perform at a human-like level across a broad spectrum of intellectual tasks. Whether Andreessen's statement is overly optimistic or insightful is open to debate, but investors are keenly wondering where AGI could first yield economic advantages.

Some believe that cryptocurrency might be a primary beneficiary. Traditional finance is burdened with know-your-customer (KYC) regulations and reporting duties, which often complicate operations for autonomous software. In contrast, an AI with a cryptocurrency wallet has the ability to transact independently. Over the past year, Ethereum has made significant strides in constructing the necessary infrastructure for such scenarios, prompting an examination of the potential impacts.

Establishing a Framework for a Machine Economy

In September 2025, the Ethereum Foundation, the non-profit entity driving the development of the Ethereum network and its ecosystem, introduced a dedicated AI division called dAI. This team aims to position Ethereum as the foundational layer for autonomous agents that can transact and collaborate without the need for human oversight.

This vision came to fruition in January 2026 when Ethereum launched the ERC-8004 standard for AI agents on its mainnet. This standard includes three essential data registries focused on tracking identity, reputation, and work validation for AI agents, effectively building a trust framework for interactions with unfamiliar counterparties. As of April 8, over 37,000 agents had registered on the Ethereum network. These developments position Ethereum as a critical player in potentially launching an AI-driven economy, complete with a growing community of active agents.

Moreover, Ethereum now enables these agents to exchange funds for goods and services through a protocol known as x402, facilitating real-time micropayments without necessitating new accounts or wallets during transactions.

Ethereum's capital resources are also noteworthy; its decentralized finance (DeFi) ecosystem boasts a total value locked (TVL) of $54 billion along with $165 billion in stablecoin assets. This robust capital foundation, combined with the x402 protocol and the introduction of ERC-8004, equips Ethereum with many of the requisite elements for an agent-based economy.

However, the sophistication of these agents remains a crucial limitation, as they currently do not exhibit AGI characteristics. Consequently, human users will remain the primary demand drivers for their services, restricting the potential for significant ecosystem expansion.

Anticipating Future Growth

Turning our attention to the implications of AGI for Ethereum, particularly if Andreessen's insights hold true, it appears that agents possessing true AGI capabilities would require more financial resources compared to the simpler AI agents currently in circulation. Providing these advanced agents with financial means or allowing them to generate income would enable them to undertake larger, more complex tasks autonomously. If much of this economic activity occurs on Ethereum, it could attract additional capital to the platform, raising demand for its cryptocurrency to cover transaction costs.

If AGI has indeed been realized, Ethereum stands to gain significantly, especially if it establishes itself as the leading platform for commerce driven by intelligent agents. However, the persistent lack of robust demand, which would lead to agents actively participating in economic activities at a scale that impacts Ethereum's network use, continues to be a concern.

Thus, while Ethereum's current positioning with AI is noteworthy and merits close observation, it does not yet signal a definitive buying opportunity. If the envisioned agent economy becomes a reality through the advancement of AGI, Ethereum could see a tremendous value surge. Conversely, if this future remains out of reach, it may spare investors from overpaying for a prospect that isn't as imminent as some of the industry's leading thinkers believe.

Marc Andreesen Claims Artificial General Intelligence (AGI) Has Arrived; Implications for Ethereum Explored.

Sharp-dressed Marc Andreessen stands on a sunlit stage at dawn, bathed in warm natural light, pointing confidently at a massive floating crystalline Ethereum logo intertwined with glowing neural-network filaments, symbolizing AGI’s arrival within the blockchain; the audience in silhouette leans forward, captivated by the breakthrough.

Marc Andreesen Just Said That Artificial General Intelligence (AGI) Is Here. Here's What That Could Mean for Ethereum. - The Motley Fool

- Marc Andreessen claims AGI exists but is not evenly distributed across sectors.  
- Ethereum has launched ERC-8004 and x402 protocols to support AI-based economic transactions.  
- The lack of sophisticated AGI limits current demand and growth potential for Ethereum.  

On April 5, Marc Andreessen, co-founder of the famous venture capital firm Andreessen Horowitz, also known as a16z, declared on social media that "AGI is already here -- it's just not evenly distributed yet." AGI, or artificial general intelligence, refers to AI systems that are capable of matching human-level performance across a wide range of intellectual tasks. It's unclear whether his claim is premature or prophetic, but for investors, the main question is where an economic benefit from AGI would occur first.

There's an argument that cryptocurrency is one such place. Traditional financial venues impose know-your-customer (KYC) regulations and reporting requirements that create friction that autonomous software can't easily navigate, whereas an artificial intelligence agent with a crypto wallet can transact without permission. Ethereum (ETH +8.20%) spent the past year building infrastructure for exactly that scenario, so let's take a look at what the implications for it could be.

Laying rails for a machine economy

The Ethereum Foundation, the nonprofit governing the network's development and its ecosystem, created a dedicated AI team called dAI in September 2025, seeking to make Ethereum the settlement layer for autonomous agents that pay and collaborate without any human intermediaries.

That ambition produced a concrete deliverable in January 2026, when the chain's new standard for AI agents, ERC-8004, went live on mainnet. ERC-8004 creates three data registries that track identity, reputation, and work validation criteria for agents, thereby creating a trust layer for AI that's transacting with unfamiliar counterparties. As of April 8, more than 37,000 agents were registered on Ethereum. So now Ethereum has a key part of the technical infrastructure needed to make an AI-based on-chain economy possible, and it also has a decent population of agents operating.

It also has the means to allow agents to actually exchange funds for services or goods. A protocol called x402 enables micropayments between agents in real time without the need to create any new accounts or wallets at the point of sale.

Expand CRYPTO : ETH Ethereum Today's Change ( 8.20 %) $ 181.29 Current Price $ 2392.11 Key Data Points Market Cap $289B Day's Range $ 2203.42 - $ 2393.44 52wk Range $ 1551.41 - $ 4946.05 Volume 29B

Furthermore, Ethereum brings a deep pool of on-chain capital for agents to tap. Its decentralized finance (DeFi) ecosystem holds $54 billion in total value locked (TVL), and $165 billion in stablecoin capital.

Between that base of capital, x402, and ERC-8004, the network thus has nearly all of the ingredients needed for a basic agent economy today. In principle, that makes the coin a bit more valuable than it would be otherwise.

The main missing ingredient is the sophistication of the agents themselves, which do not currently meet the criteria for being or having AGI. Without it, the main source of demand for their capabilities will come from human users, and the bulk of the goods and services that will be supplied to the market will be the result of a human's design and deployment, which limits the ecosystem's growth potential.

There could be plenty of growth on the way

Now, let's turn to the question of what artificial general intelligence would mean for Ethereum, specifically given its positioning, and assuming that Andreessen is correct.

Agents with actual AGI-level capabilities would likely have a larger need for financial resources than the less-complex AI agents that we've seen in operation so far. Equipping the stronger agents with funds, or allowing them to earn their own money, would be helpful because it'd enable them to purchase more services and take on larger and larger tasks with a higher level of independence than they'd be capable of otherwise. If a lot of the economic activity associated with those powerful agents occurs on Ethereum, it'd likely attract a lot more capital to the chain, and boost demand for the coin somewhat, as it's necessary to pay for transaction costs.

So if AGI really has been invented, there's likely some upside in store, and potentially a lot if Ethereum becomes the dominant platform for agent-based commerce. But again, for the moment, what's missing is sources of demand that would lead to actual agents doing economic work at a scale that moves the needle for Ethereum's network usage.

Thus, Ethereum's AI positioning is a thesis worth tracking rather than a buy signal today. If the agent economy materializes thanks to the development of AGI, it could capture enormous value. If it doesn't, you'll have avoided paying a premium for a future that was further away than what some of the smartest thinkers assumed.

The realm of artificial intelligence (AI) is intricate and often laden with specialized terminology that can make it difficult to comprehend. To bridge this gap, we have compiled a comprehensive glossary offering clear definitions of key terms and phrases pertinent to AI, which we will continuously update as advancements and safety considerations emerge in the field.

One term frequently discussed is artificial general intelligence (AGI), which refers to AI systems that can outperform humans across a wide variety of tasks. OpenAI CEO Sam Altman recently likened AGI to an average human colleague, while OpenAI's charter describes it as highly autonomous systems exceeding human capabilities in valuable economic work. In contrast, Google DeepMind interprets AGI as AI matching human performance in most cognitive areas. Unsurprisingly, even experts in AI research find these definitions somewhat ambiguous.

An AI agent denotes a sophisticated tool that utilizes AI capabilities to execute complex tasks beyond the scope of basic AI chatbots. This could encompass actions like submitting expense reports, reserving restaurant tables, or managing code. The term can vary in meaning depending on the user's context, and the necessary infrastructure to support such agents is still evolving. In essence, an AI agent represents an autonomous system that leverages various AI solutions to accomplish multiple-step tasks.

A prime example of AI's capabilities is seen in the concept of chain-of-thought reasoning. This process allows large language models to deconstruct problems into smaller, manageable parts, leading to more accurate results, especially in logical or programming scenarios. Though this method may take longer, it often enhances overall accuracy. Models optimized for such reasoning typically draw on traditional large language models and employ reinforcement learning techniques.

In the world of AI, "compute" generally refers to the essential computational power that enables AI models to function. This power is fundamentally supported by various hardware components such as GPUs, CPUs, TPUs, and essential infrastructure that drives the modern AI landscape.

Deep learning, a subset of machine learning, utilizes multi-layered artificial neural networks (ANNs) to identify complex patterns in data. Unlike simpler machine learning approaches, deep learning can autonomously discern significant features in datasets, continuously improving its output through iterative learning. Nonetheless, these models demand vast amounts of data—often millions of examples—to achieve optimal performance and typically involve higher training times and costs.

The diffusion process plays a crucial role in generating AI models for art, music, and text. It mimics physical diffusion by gradually "destroying" data structures through noise addition. However, AI systems aim to learn a reversed version of this process to successfully recover the original data.

Distillation is another valuable technique in AI development, where knowledge is extracted from a larger AI model (referred to as the "teacher") to create a smaller, more efficient version (the "student"). This process enhances performance while minimizing loss in effectiveness, a method that likely contributed to the development of OpenAI's GPT-4 Turbo.

Fine-tuning focuses on refining an AI model to maximize performance for specific tasks by introducing specialized, task-oriented data to the existing training framework. Numerous AI startups utilize large language models as foundations, seeking to enhance utility for particular sectors by augmenting initial training with domain expertise.

Generative Adversarial Networks (GANs) are vital in generative AI, utilizing two neural networks engaged in a competitive process. One network generates data outputs while the second assesses them, creating a feedback loop that improves the realism of the generated data over time. Although effective for specific applications, GANs may not be suitable for broad, general-purpose use.

The term "hallucination" in AI refers to models producing false or misleading information, posing significant challenges to AI reliability. These inaccuracies stem from limitations in training data, leading to a rising interest in developing specialized AI models designed to mitigate such issues.

Inference represents the application of an AI model to derive insights from previously examined data, fundamentally reliant on prior training. Various hardware configurations can execute inference, although performance can differ markedly based on the capabilities of the devices involved.

Large language models, such as OpenAI's GPT and competing platforms, are the backbone of popular AI assistants. These models process requests by evaluating millions of text patterns to deliver contextually relevant responses.

The memory cache is a critical element that enhances inference efficiency, optimizing how an AI system generates responses. By storing frequently used calculations, caching reduces computing demands and accelerates answer generation. Different caching mechanisms exist, with key-value (KV) caching being a notable example in transformer-based models.

Neural networks provide the foundational architecture for deep learning, drawing inspiration from the intricate interconnections of the human brain. Recent advancements in graphical processing hardware have significantly empowered the efficacy of these networks by enabling more layers and enhancing performance across diverse applications.

"RAMageddon" describes a growing crisis in the tech sector marked by an acute shortage of RAM chips, essential for running technology products. The surge in AI applications has led major tech firms to monopolize RAM for their data centers, driving prices up and impacting various industries, from gaming to consumer electronics.

The training process in machine learning entails feeding data into models so they can recognize patterns and deliver useful outputs. It is pivotal for developing adaptive AI systems, unlike rules-based AIs, which follow predetermined instructions without iterative learning. Training can be resource-intensive and costly, prompting some developers to adopt hybrid strategies that reduce training demands.

Tokens form the core of human-machine interactions, representing discrete segments of data that facilitate communication between users and AI systems. Tokenization breaks down language inputs into manageable units, making them comprehensible to AI. Various types of tokens exist, such as input and output tokens, and the volume of token usage often influences costs in enterprise AI applications.

Transfer learning allows developers to leverage previously trained models as a base for new ones, facilitating more efficient development, especially when limited data is available for the new task. Nonetheless, models relying on this technique will likely require additional training to refine their domain-specific capabilities.

Lastly, weights serve as crucial elements in AI model training, determining the significance of various input features and directly influencing model outputs. Initially set randomly, these numerical parameters adjust throughout the training process, refining the model's predictive accuracy based on the provided data.

This article will continue to evolve as new developments emerge in the AI sphere.

A straightforward guide to common AI terminology, from LLMs to hallucinations.

A cozy wooden desk bathed in warm morning sunlight, an open tome titled “AI Glossary” at its center. Shimmering words—LLM, token, hallucination, neural network—rise like sparks off the pages, each term morphing into a tiny illustrative icon (a brain grid, a glitching lightbulb, intertwined nodes). A lone adult figure’s silhouette leans in from the side, pen in hand, as if jotting down notes. The room’s walls are lined with shelves of books and a steaming mug of coffee sits nearby, evoking a calm, enlightening moment of discovery.

From LLMs to hallucinations, here’s a simple guide to common AI terms - TechCrunch

- A glossary has been created to clarify AI terminology for better industry understanding.  
- Artificial general intelligence (AGI) refers to AI outperforming humans at economically valuable work.  
- Chain-of-thought reasoning enhances AI problem-solving by breaking down tasks into smaller steps.  

Artificial intelligence is a deep and convoluted world. The scientists who work in this field often rely on jargon and lingo to explain what they’re working on. As a result, we frequently have to use those technical terms in our coverage of the artificial intelligence industry. That’s why we thought it would be helpful to put together a glossary with definitions of some of the most important words and phrases that we use in our articles.

We will regularly update this glossary to add new entries as researchers continually uncover novel methods to push the frontier of artificial intelligence while identifying emerging safety risks.

Artificial general intelligence, or AGI, is a nebulous term. But it generally refers to AI that’s more capable than the average human at many, if not most, tasks. OpenAI CEO Sam Altman recently described AGI as the “equivalent of a median human that you could hire as a co-worker.” Meanwhile, OpenAI’s charter defines AGI as “highly autonomous systems that outperform humans at most economically valuable work.” Google DeepMind’s understanding differs slightly from these two definitions; the lab views AGI as “AI that’s at least as capable as humans at most cognitive tasks.” Confused? Not to worry — so are experts at the forefront of AI research.

An AI agent refers to a tool that uses AI technologies to perform a series of tasks on your behalf — beyond what a more basic AI chatbot could do — such as filing expenses, booking tickets or a table at a restaurant, or even writing and maintaining code. However, as we’ve explained before, there are lots of moving pieces in this emergent space, so “AI agent” might mean different things to different people. Infrastructure is also still being built out to deliver on its envisaged capabilities. But the basic concept implies an autonomous system that may draw on multiple AI systems to carry out multistep tasks.

Given a simple question, a human brain can answer without even thinking too much about it — things like “which animal is taller, a giraffe or a cat?” But in many cases, you often need a pen and paper to come up with the right answer because there are intermediary steps. For instance, if a farmer has chickens and cows, and together they have 40 heads and 120 legs, you might need to write down a simple equation to come up with the answer (20 chickens and 20 cows).

In an AI context, chain-of-thought reasoning for large language models means breaking down a problem into smaller, intermediate steps to improve the quality of the end result. It usually takes longer to get an answer, but the answer is more likely to be correct, especially in a logic or coding context. Reasoning models are developed from traditional large language models and optimized for chain-of-thought thinking thanks to reinforcement learning.

(See: Large language model)

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

Although somewhat of a multivalent term, compute generally refers to the vital computational power that allows AI models to operate. This type of processing fuels the AI industry, giving it the ability to train and deploy its powerful models. The term is often a shorthand for the kinds of hardware that provides the computational power — things like GPUs, CPUs, TPUs, and other forms of infrastructure that form the bedrock of the modern AI industry.

A subset of self-improving machine learning in which AI algorithms are designed with a multi-layered, artificial neural network (ANN) structure. This allows them to make more complex correlations compared to simpler machine learning-based systems, such as linear models or decision trees. The structure of deep learning algorithms draws inspiration from the interconnected pathways of neurons in the human brain.

Deep learning AI models are able to identify important characteristics in data themselves, rather than requiring human engineers to define these features. The structure also supports algorithms that can learn from errors and, through a process of repetition and adjustment, improve their own outputs. However, deep learning systems require a lot of data points to yield good results (millions or more). They also typically take longer to train compared to simpler machine learning algorithms — so development costs tend to be higher.

(See: Neural network)

Diffusion is the tech at the heart of many art-, music-, and text-generating AI models. Inspired by physics, diffusion systems slowly “destroy” the structure of data — for example, photos, songs, and so on — by adding noise until there’s nothing left. In physics, diffusion is spontaneous and irreversible — sugar diffused in coffee can’t be restored to cube form. But diffusion systems in AI aim to learn a sort of “reverse diffusion” process to restore the destroyed data, gaining the ability to recover the data from noise.

Distillation is a technique used to extract knowledge from a large AI model with a ‘teacher-student’ model. Developers send requests to a teacher model and record the outputs. Answers are sometimes compared with a dataset to see how accurate they are. These outputs are then used to train the student model, which is trained to approximate the teacher’s behavior.

Distillation can be used to create a smaller, more efficient model based on a larger model with a minimal distillation loss. This is likely how OpenAI developed GPT-4 Turbo, a faster version of GPT-4.

While all AI companies use distillation internally, it may have also been used by some AI companies to catch up with frontier models. Distillation from a competitor usually violates the terms of service of AI API and chat assistants.

This refers to the further training of an AI model to optimize performance for a more specific task or area than was previously a focal point of its training — typically by feeding in new, specialized (i.e., task-oriented) data.

Many AI startups are taking large language models as a starting point to build a commercial product but are vying to amp up utility for a target sector or task by supplementing earlier training cycles with fine-tuning based on their own domain-specific knowledge and expertise.

(See: Large language model [LLM])

A GAN, or Generative Adversarial Network, is a type of machine learning framework that underpins some important developments in generative AI when it comes to producing realistic data – including (but not only) deepfake tools. GANs involve the use of a pair of neural networks, one of which draws on its training data to generate an output that is passed to the other model to evaluate. This second, discriminator model thus plays the role of a classifier on the generator’s output – enabling it to improve over time.

The GAN structure is set up as a competition (hence “adversarial”) – with the two models essentially programmed to try to outdo each other: the generator is trying to get its output past the discriminator, while the discriminator is working to spot artificially generated data. This structured contest can optimize AI outputs to be more realistic without the need for additional human intervention. Though GANs work best for narrower applications (such as producing realistic photos or videos), rather than general purpose AI.

Hallucination is the AI industry’s preferred term for AI models making stuff up – literally generating information that is incorrect. Obviously, it’s a huge problem for AI quality.

Hallucinations produce GenAI outputs that can be misleading and could even lead to real-life risks — with potentially dangerous consequences (think of a health query that returns harmful medical advice). This is why most GenAI tools’ small print now warns users to verify AI-generated answers, even though such disclaimers are usually far less prominent than the information the tools dispense at the touch of a button.

The problem of AIs fabricating information is thought to arise as a consequence of gaps in training data. For general purpose GenAI especially — also sometimes known as foundation models — this looks difficult to resolve. There is simply not enough data in existence to train AI models to comprehensively resolve all the questions we could possibly ask. TL;DR: we haven’t invented God (yet).

Hallucinations are contributing to a push towards increasingly specialized and/or vertical AI models — i.e. domain-specific AIs that require narrower expertise – as a way to reduce the likelihood of knowledge gaps and shrink disinformation risks.

Inference is the process of running an AI model. It’s setting a model loose to make predictions or draw conclusions from previously seen data. To be clear, inference can’t happen without training; a model must learn patterns in a set of data before it can effectively extrapolate from this training data.

Many types of hardware can perform inference, ranging from smartphone processors to beefy GPUs to custom-designed AI accelerators. But not all of them can run models equally well. Very large models would take ages to make predictions on, say, a laptop versus a cloud server with high-end AI chips.

[See: Training]

Large language models, or LLMs, are the AI models used by popular AI assistants, such as ChatGPT, Claude, Google’s Gemini, Meta’s AI Llama, Microsoft Copilot, or Mistral’s Le Chat. When you chat with an AI assistant, you interact with a large language model that processes your request directly or with the help of different available tools, such as web browsing or code interpreters.

AI assistants and LLMs can have different names. For instance, GPT is OpenAI’s large language model and ChatGPT is the AI assistant product.

LLMs are deep neural networks made of billions of numerical parameters (or weights, see below) that learn the relationships between words and phrases and create a representation of language, a sort of multidimensional map of words.

These models are created from encoding the patterns they find in billions of books, articles, and transcripts. When you prompt an LLM, the model generates the most likely pattern that fits the prompt. It then evaluates the most probable next word after the last one based on what was said before. Repeat, repeat, and repeat.

(See: Neural network)

Memory cache refers to an important process that boosts inference (which is the process by which AI works to generate a response to a user’s query). In essence, caching is an optimization technique, designed to make inference more efficient. AI is obviously driven by high-octane mathematical calculations and every time those calculations are made, they use up more power. Caching is designed to cut down on the number of calculations a model might have to run by saving particular calculations for future user queries and operations. There are different kinds of memory caching, although one of the more well-known is KV (or key value) caching. KV caching works in transformer-based models, and increases efficiency, driving faster results by reducing the amount of time (and algorithmic labor) it takes to generate answers to user questions.

(See: Inference)

A neural network refers to the multi-layered algorithmic structure that underpins deep learning — and, more broadly, the whole boom in generative AI tools following the emergence of large language models.

Although the idea of taking inspiration from the densely interconnected pathways of the human brain as a design structure for data processing algorithms dates all the way back to the 1940s, it was the much more recent rise of graphical processing hardware (GPUs) — via the video game industry — that really unlocked the power of this theory. These chips proved well suited to training algorithms with many more layers than was possible in earlier epochs — enabling neural network-based AI systems to achieve far better performance across many domains, including voice recognition, autonomous navigation, and drug discovery.

(See: Large language model [LLM])

RAMageddon is the fun new term for a not-so-fun trend that is sweeping the tech industry: an ever-increasing shortage of random access memory, or RAM chips, which power pretty much all the tech products we use in our daily lives. As the AI industry has blossomed, the biggest tech companies and AI labs — all vying to have the most powerful and efficient AI — are buying so much RAM to power their data centers that there’s not much left for the rest of us. And that supply bottleneck means that what’s left is getting more and more expensive.

That includes industries like gaming (where major companies have had to raise prices on consoles because it’s harder to find memory chips for their devices), consumer electronics (where memory shortage could cause the biggest dip in smartphone shipments in more than a decade), and general enterprise computing (because those companies can’t get enough RAM for their own data centers). The surge in prices is only expected to stop after the dreaded shortage ends but, unfortunately, there’s not really much of a sign that’s going to happen anytime soon.

Developing machine learning AIs involves a process known as training. In simple terms, this refers to data being fed in in order that the model can learn from patterns and generate useful outputs.

Things can get a bit philosophical at this point in the AI stack — since, pre-training, the mathematical structure that’s used as the starting point for developing a learning system is just a bunch of layers and random numbers. It’s only through training that the AI model really takes shape. Essentially, it’s the process of the system responding to characteristics in the data that enables it to adapt outputs towards a sought-for goal — whether that’s identifying images of cats or producing a haiku on demand.

It’s important to note that not all AI requires training. Rules-based AIs that are programmed to follow manually predefined instructions — for example, such as linear chatbots — don’t need to undergo training. However, such AI systems are likely to be more constrained than (well-trained) self-learning systems.

Still, training can be expensive because it requires lots of inputs — and, typically, the volumes of inputs required for such models have been trending upwards.

Hybrid approaches can sometimes be used to shortcut model development and help manage costs. Such as doing data-driven fine-tuning of a rules-based AI — meaning development requires less data, compute, energy, and algorithmic complexity than if the developer had started building from scratch.

[See: Inference]

When it comes to human-machine communication, there are some obvious challenges. People communicate using human language, while AI programs execute tasks and respond to queries through complex algorithmic processes that are informed by data. In their simplest definition, tokens represent the basic building blocks of human-AI communication, in that they are discrete segments of data that have either been processed or produced by an LLM.

Tokens are created via a process known as “tokenization,” which breaks down raw data and refines it into distinct units that are digestible to an LLM. Similar to how a software compiler translates human language into binary code that a computer can digest, tokenization interprets human language for an AI program via their user queries so that it can prepare a response.

There are several different kinds of tokens — including input tokens (the kind that must be generated in response to a human user’s query), output tokens (the kind that are generated as the LLM responds to the human’s request), and reasoning tokens, which involve longer, more intensive tasks and processes that occur as part of a user request.

With enterprise AI, token usage also determines costs. Since tokens are equivalent to the amount of data being processed by a model, they have also become the means by which the AI industry monetizes its services. Most AI companies charge for LLM usage on a per-token-basis. Thus, the more tokens a business burns as it uses an AI program (ChatGPT, for example), the more money it will have to pay its AI service provider (OpenAI).

A technique where a previously trained AI model is used as the starting point for developing a new model for a different but typically related task – allowing knowledge gained in previous training cycles to be reapplied.

Transfer learning can drive efficiency savings by shortcutting model development. It can also be useful when data for the task that the model is being developed for is somewhat limited. But it’s important to note that the approach has limitations. Models that rely on transfer learning to gain generalized capabilities will likely require training on additional data in order to perform well in their domain of focus

(See: Fine tuning)

Weights are core to AI training, as they determine how much importance (or weight) is given to different features (or input variables) in the data used for training the system — thereby shaping the AI model’s output.

Put another way, weights are numerical parameters that define what’s most salient in a dataset for the given training task. They achieve their function by applying multiplication to inputs. Model training typically begins with weights that are randomly assigned, but as the process unfolds, the weights adjust as the model seeks to arrive at an output that more closely matches the target.

For example, an AI model for predicting housing prices that’s trained on historical real estate data for a target location could include weights for features such as the number of bedrooms and bathrooms, whether a property is detached or semi-detached, whether it has parking, a garage, and so on.

Ultimately, the weights the model attaches to each of these inputs reflect how much they influence the value of a property, based on the given dataset.

This article is updated regularly with new information.

From LLMs to hallucinations, here’s a simple guide to common AI terms

The intense focus on artificial intelligence (AI) development by major tech players like Google Cloud and Amazon (AWS), alongside prominent AI research labs such as OpenAI, Anthropic, and DeepMind, has become increasingly apparent. The top four companies—Amazon, Google, Microsoft, and Meta—have committed over $600 billion towards AI initiatives by 2026. This growing trend has dominated the equity markets over the past three years, leading to Nvidia’s rise as the most valuable public company, largely due to its cutting-edge AI hardware. As the hype surrounding chatbots diminishes, investment may increasingly target tangible AI infrastructure such as data centers and energy management systems, along with the evolution of autonomous systems capable of independent thought and complex task execution.

The implications of AI development raise pertinent questions, especially for practicing surgeons concerning job security. Although there have been persistent claims that AI would overtake roles in radiology, these specialists are engaged in more work than ever. Recent comments from Elon Musk, suggesting that medical education is “pointless," add to the uncertainty.

To comprehend these developments, it's crucial to define artificial general intelligence (AGI). In simplified terms, AGI refers to a form of AI that matches or surpasses human capabilities across various cognitive tasks and learns new skills faster than humans can, even without targeting specific training. For context, consider Tesla’s Full Self-Driving (FSD) technology; while it exemplifies advanced artificial narrow intelligence (ANI) by executing a focused task like driving, it does not possess the versatility of AGI. A Tesla cannot autonomously transition from driving to cooking or performing dental hygiene tasks because it was engineered for a singular function.

Should AGI be realized, it would dramatically exceed Tesla’s capabilities, outperforming it and potentially excelling in various roles such as trucking, programming, and even complex surgical tasks. The discussion also addresses whether AGI would outpace human surgeons in performing surgeries.

Currently, there is a divide among leading AI researchers regarding the feasibility of achieving true AGI. Some are cautiously reassessing their focus, recognizing that the fiscal investments in creating AGI are significant, while others are diverting efforts to develop ANI solutions. These narrower AI models can replace roles in programming, legal research, and project management.

For companies like Anthropic, OpenAI, and Google, there are two strategic pathways: either monetize existing ANI capabilities or continue the pursuit of AGI, which could lead to being first to redefine all operational frontiers in AI. The first route may foster a network of highly specialized ANI agents, capable of performing at human levels in narrow domains, whereas the second posits that near-term breakthroughs in programming could render ANI obsolete as AGI arrives.

Imagining a future where multiple ANIs collaborate in surgical settings paints an intriguing picture. For example, one ANI could surpass human vascular surgeons in analyzing CT scans, while another might excel at case planning or interfacing with hospital scheduling systems. However, while we may see some AI-supported functions materialize in surgical environments soon, fully integrated robotic systems may still be a decade away, bound by technical and regulatory challenges.

The contrast in views asserts that AGI could revolutionize healthcare within an astonishingly brief timeframe. Such developments could allow AGI-operated robots to handle entire patient care pathways, vastly outperforming their human counterparts across all roles in hospitals.

However, skepticism remains regarding the technical underpinnings necessary for AGI. Significant hurdles must be overcome, particularly in math and computer architecture, to advance the current AI frameworks. Efforts are underway to seek alternative approaches to improve model performance beyond just scaling up.

Recent market trends suggest a dramatic reassessment of the AI landscape, with significant declines in software company valuations and shifts in investor sentiments towards AI risk. Companies like Anthropic have initiated ANI models that highlight a looming threat for software products, which has contributed to a re-evaluation of future profitability.

In the context of careers, particularly for surgeons, there’s speculation regarding surgical positions potentially being affected. HALO stocks—those showcasing heavy assets and low obsolescence—stand out as more stable against these transitions. However, even surgical roles may face challenges, especially if AI can replace significant human labor.

Preparing for this transformative AI landscape may require surgeons to adapt by focusing on complex procedural competences that are less likely to be automated soon. Learning about AI is also advisable to better understand and navigate this evolving field.

Despite the uncertainties surrounding AI's future role in medicine, there is optimism about the potential for it to enhance efficiency and patient outcomes in the immediate term. However, a longer-term outlook indicates profound changes ahead—signaling the last generation of human surgeons may well be in training now. 

Written by David Stonko, a vascular surgery fellow.

The future of surgery and artificial general intelligence

A bright, sunlit futuristic operating room where a sleek humanoid surgical robot with articulated arms patterned in glowing neural-network lines gently hands a precision scalpel to a human surgeon; large windows cast warm natural light across stainless steel instruments and hologram-free digital displays showing 3D anatomical maps; a team of focused medical professionals watches as man and machine collaborate on a breakthrough procedure, evoking the dawn of AGI-powered surgery.

Artificial general intelligence and the future of surgery - KevinMD.com

- Major tech companies have invested over $600 billion in AI development by 2026.
- Hyperscalers and AI labs face a divide on pursuing AGI versus developing narrow AI solutions.
- Surgeons may be at risk of obsolescence from advancing AI technologies in the future.

Ben Goertzel, dubbed the 'Father of AGI,' claims human-level AI is 2-3 years from realization and issues a warning about the implications.

More from Artificial General Intelligence

Prepare for an AI-driven future with these 3 skills — AGI pioneer warns that ‘human jobs will be outdated’

Why Google DeepMind Recently Dropped Single-Score AI Evaluations

Transitioning from AI to AGI

What is ‘Jagged Intelligence’ and how can it reshape the conversation around AI?