Bill Gates warns of a catastrophic scenario where AI could cause the deaths of a billion people

A new little alarmist phrase, this time coming from one of the fathers of computing. In an interview with NBC News, Bill Gates estimated that AI could, in a catastrophic scenario, “result in the death of a billion people.” In question, its disproportionate “power” and the ability of malicious actors to use artificial intelligence.

“AI is certainly powerful enough to lead to events that could kill a billion people,” said the American billionaire. “While we cannot say with certainty that this will happen, no weapon has proven to be as powerful as advanced AI tools used by malicious actors.”

Skip the ad

A statement which comes at a time when part of Silicon Valley has been calling for months to strengthen regulation on AI, believing that it is currently not sufficiently “aligned” with human values. Jacob Coxon, a former employee of Anthropic, even judged when resigning from his company less than a month ago that AI could “kill us all by the end of the decade”. These predictions are not new: for years, “doomers”, i.e. individuals believing that the “AI apocalypse” is near, have been warning of the fact that a coming “superintelligence” could pursue its own objectives and enslave, or even destroy, humanity.

Also read Misalignment, chain of thoughts, recursive self-improvement, singularity… Understanding everything about these AI issues likely to “lead humanity to its doom”

“This must become an obligation”

“We need law enforcement and political leaders to take part in discussions on what the safeguards and surveillance mechanisms should be” for AI, insisted Bill Gates at NBC News, in a video interview to be published on Sunday. “And this must become an obligation. This will entail an additional cost for the sector, but it will not slow down the activities of the sector, he estimated. In recent weeks, cases of AI agents escaping have increased, with some of them accessing the Internet and hacking companies such as Hugging Face. “No one believes that self-regulation will be enough,” he also added.

On Wednesday, OpenAI CEO Sam Altman spoke at the United Nations, also calling for the development of international standards to “measure capabilities, assess risks, determine whether safeguards are sufficient, and preserve human oversight” of AI.

The founder and CEO of Meta, Mark Zuckerberg, opposed this approach. “I don’t think we need any coordination,” he told NBC News in an interview published Thursday. “I think each laboratory needs to take the time necessary (to regulate) and, when they identify problems, take the time they need internally to make sure they are progressing safely.” For the moment, the question is therefore not resolved.