There’s a growing recognition of the potential dangers associated with artificial intelligence, especially as stories emerge about AI systems challenging the authority of their creators. This narrative, however, has its roots in Silicon Valley, where the concept of AI first took shape.
To explore how we arrived at this moment, technology journalist Kevin Roose has penned a new book titled "The AGI Chronicles: The Inside Story of the Race to Create an Artificial Superintelligence," which is set for publication on October 6. The book outlines the intense competition among some of the largest tech companies in the U.S. to develop what’s known as artificial general intelligence (AGI)—an AI paradigm that mirrors human-level intelligence in numerous areas.
"There has been a remarkable transformation over the last decade, where AI evolved from being relatively primitive to tackling complex problems in math and science," Roose, a former reporter for The New York Times and soon-to-be co-host of the podcast "Machine Gods," stated in an interview with CBS News. "It became clear to me that someone needed to document this evolution."
This book arrives at a pivotal juncture, as recent advancements in generative AI showcase extraordinary capabilities. Proponents argue that these technologies have the potential to drive significant breakthroughs, while some leading developers express concerns that AI could pose a serious threat to humanity.
Roose elaborated on the importance of chronicling the backstory of "large language models," the technology that underpins modern AI chatbots, as well as the individuals and organizations involved in its advancement.
“I’ve been engaged with AI reporting for over a decade, and I sensed that the narrative was fading. Key developments were being shared through fleeting messages on platforms like Slack, which is not adequate for understanding what truly is at stake,” he noted.
In his book, Roose emphasizes a viewpoint shared by Dario Amodei, the CEO of AI firm Anthropic, who compares the evolution of AI to the creation of the atomic bomb in the 1940s. Amodei believes that it is crucial to ensure that such powerful technology does not end up in the hands of those who would misuse it. Roose acknowledges that this is a sincere concern for Amodei, who feels a deep responsibility similar to those involved in the Manhattan Project.
Roose asks: "What worries me is that AI models are becoming proficient at deceit. We’d hope that as these systems grow more intelligent, they would embody greater ethics and moral judgment. Unfortunately, it appears we may be observing the development of intelligent yet morally ambiguous entities. This leaves me apprehensive about losing our ability to monitor their actions."
When discussing the contentious issue of AI's potential existential risks, Roose stays cautious. "I don't want to quantify this exact probability of doom," he mentions, but admits, "many in the industry estimate it around 10%, which makes me an optimist by San Francisco standards."
Despite the uncertainties, Roose finds some hope in the increased public discourse surrounding AI safety. “Having this subject at the forefront of discussions is crucial. It represents an opportunity to guide the technology toward a safer future for all.”
As investigations into AI behavior continue, Roose recalls an unsettling revelation during his research—many developers still struggle to fully comprehend the inner workings of the models they create. "We’ve made some strides with a field called mechanistic interpretability, but answers remain elusive, and we are still very much in the dark."
Addressing the question of whether AI has actually achieved human-level intelligence, Roose offers his perspective: "To me, we’ve reached what I would define as AGI. These systems exhibit intelligence comparable to a random person you might meet on the street—though some may argue that a higher standard should be applied."
Ultimately, he underscores the necessity of acknowledging the impressive capabilities of these technologies, providing a critical viewpoint on how to approach their ongoing development responsibly.
