Chief AI Leaders Issue Warnings on Developing Tech

A researcher at Anthropic announced Sept. 8 that he was leaving his position and issued an urgent warning that artificial intelligence could “kill us all by the end of the decade.” This researcher, 27-year-old Jason Coxon, condemned Anthropic and OpenAI, the leading companies in AI, for acting irresponsibly. He shared his belief that only a collective shutdown of the industry or severe government intervention can safely move the technology forward. 

Later, Anthropic’s CEO released a milder but similar statement calling for a slower development pace and closer monitoring. Dario Amodei published his online essay, titled We Must Pace the Frontier, detailing a three-point plan for further study. The most supported idea is the need for third-party safety monitors. OpenAI leader Sam Altman and xAI’s Elon Musk both agreed with Amodei’s assessment of the serious risks of further development. 

Altman revealed that his company has recently discussed similar concerns: “This has been a primary topic of discussion we’ve had at OpenAI in recent weeks,” he shared on X. Musk simply shared, “Dario is right.” This kind of broad agreement is rare in the AI industry, as companies have previously been locked in stark competition with one another. 

However, these warnings have received some backlash, specifically from AI development company Nvidia’s CEO. “The market forces are already there; we don’t need any new laws, we don’t need new regulations,” said Jenson Huang. 

Meta’s Mark Zuckerberg also emphasized the importance of handling the issue individually. Neither of the pair finds coordination necessary. “If you build a product or a service, and you’re not confident in its functionality, capability, or safety, then don’t release it,” stated Huang at Salesforce’s Dreamforce Conference in San Francisco. 

In a piece by The Guardian, several AI experts assessed the probability of the six most threatening claims made about this technology in the recent weeks. The most prominent claim was made by Anthropic’s Amodei, who warned that a swarm of intelligent tech could take over the entirety of the internet, creating what he describes as a “persistent botnet”. If this were the case, all information being input and output from the World Wide Web would be controlled by artificial intelligence. New York University professor Gary Marcus disputed the claim as “nonsensical” because Anthropic’s current utilities can only overtake smaller, vulnerable websites. Sites like Google and Amazon Web Services are unlikely to fall as they make up a larger percentage of the internet and are well defended.  

One of the most dangerous threats of a super-intelligent AI is its potential access to nuclear weaponry. “Just as we can’t uninvent the bomb, once we develop AI systems that surpass humanity, it would be almost impossible to go back,” said Oxford senior researcher Toby Ord. However, an AI of this capacity has not yet been developed. A similar concern is that AI could use disinformation to pressure the use of nuclear weapons by humans, which is why the United Nations determined that oversight and control of nuclear weapon systems should remain solely human. 

• #AI
• college
• debates
• growing AI
• industry
• insta
• meta
• Musk
• warning
• zuckerberg