You’ve heard that AI could kill us all. Yes, so could old age.
At Fast Company, we learn that “Bill Gates has a dire warning about AI,” publicized at the Telluride Film Festival. He prophesied “mass unemployment and major social, political, and economic upheaval across the globe.”
His solution? In a 6,000-word essay,
Gates also proposed major limits on AI development, arguing that “a public democratic process that includes elected officials, policymakers, educators, health workers, local officials, and community leaders” should determine future solutions, not the AI companies themselves.
His suggested approach would end up giving Silicon Valley the powers of government to cope with an emergency that the Valley itself created.
So the tech media are helping to spread a highly convenient panic:
- “The AI Researcher Who Just Quit Anthropic Says It’s ‘Crunch Time for Humanity’” (Wired)
- “Why So Many AI Researchers Think the Machines Could Kill Everyone” (Wired)
- “‘The People Building AI Earnestly Believe That It Could Kill Us All’: Anthropic Researcher Quits Dramatically” (Gizmodo)
That’s no surprise. All this drama inflates the apparent importance of tech media as well.
Okay, First, Who Quit and Why?
Anthropic researcher Jacob Coxon quit and posted at X last Tuesday, to over 25 million views, that “No other human activity poses this level of danger.”
Not even nuclear war? From Wired:
Genuinely stunning advances in AI capabilities — an OpenAI model solved a centuries-old math problem in a matter of hours — have come amid a rash of security incidents that saw swarms of agents break free from containment to hack into other systems.
Those concerns reached a fever pitch this week after researcher Jacob Coxon announced his resignation from Anthropic while warning that AI firms are “racing straight to self-improving superintelligence and gambling with our lives.” A senior Anthropic leader — who works on AI safety — piped up with a similarly blunt assessment: “We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade.”
Actually, the jury is still out about that big math advance. The breach of security hacking incidents suggest that the entire industry needs to get its act together, perhaps with a bit of prodding from law enforcement and other authorities.
But these things have happened to the nuclear industry from time to time too. Yet the Atomic Doomsday Clock has not reached midnight since it was first developed in 1947.
That Doesn’t Mean We Should Ignore the Threat
Twenty-five years ago today, the United States paid dearly for ignoring the threat of Islamic terror.
At the same time, we need to see that insiders’ panic is often wrapped up in their personal stake in the problems and the possible solutions. So it’s best to step back and assess those questions carefully.
For example, in response to one media doomsday scenario, AI analyst Gary Marcus offered a too-rare rebuke, “No, Anderson Cooper, AI is not going to kill all humans by 2030. AI is already causing harms, there are are many real risks, but constant focus on absurd end-of-the-world fantasies has made the situation worse”:
Indeed, there is good reason to doubt that so-called Doomers have any clue what they are doing in terms of enacting change in the real world, or even in thinking through the consequences of their own actions. They are fabulous at getting publicity… And they are fabulous at getting truly massive funding, perhaps close to (if not over) a billion dollars by now.
And, if Bill Gates is any example, some of the Doomers aim at legislative power too. Without having the least idea how politics works outside their labs.
A Way Forward?
Marcus points to two recent constructive developments: Senator Blumenthal has sent a letter to Sam Altman at OpenAI, demanding some answers about recent security lapses, particularly the Hugging Face incident. Second, Protect Democracy, a public interest law firm, has filed a lawsuit,demanding more openness from the federal government about the criteria by which AI developments are evaluated.
Meanwhile, Jay Richards, pointing to Congressman Ro Khanna — who seems to be inflating the panic — warns, “Take no one seriously who claims they can calculate a probability of ‘AI causing human extinction.’ Model users can’t accurately predict the future temperature with a problem that is at least well-defined. ‘AI causing human extinction’ is a vague Nostradamus-style ‘prophecy.’”
Citizens starting to access their right to know — and exercising much more skepticism — will likely prove far more useful in dealing with AI than prophesying apocalypses or giving the AI firms political power.
Cross-posted at Mind Matters News.