Some people oppose regulating AI, arguing it would slow down innovation.
Hinton's response: "They want a speeding sports car without a steering wheel."
He made the remark on April 21, 2026, at the Digital World 2026 conference in Geneva. Hinton appeared via video link — 78 years old, winner of the 2024 Nobel Prize in Physics, and three years removed from his departure from Google, which he left because he believed the risks of AI were too serious and he didn't want to hold back his views while still employed.
"Only 1% of work focuses on safety"
He offered a figure: about 1% of AI work focuses on safety.
The remaining 99%? It's about making AI run faster, smarter, and cheaper.
"It's crazy," Hinton said.
Not a nuanced academic statement — just a direct declaration: it's crazy.
Public team ratios from major AI labs show safety researchers to capabilities researchers at roughly 1:10 to 1:20. Anthropic has done relatively better on this ratio, but the industry-wide resource tilt is obvious.
What has already happened, and what hasn't
Call center jobs are already being replaced by AI. This is not a prediction; it is a reality unfolding now.
Hinton's forecast went a step further:
"If we get superintelligent AI, it can do any intellectual work."
There is a key word in that sentence: any. Writing code, drafting legal documents, doing financial analysis, making medical diagnoses — all fall under that "any." His point is not that "some jobs will disappear," but that "the entire economic foundation of intellectual labor will be upended."
He did not give a precise timeline. That is consistent with his approach — he does not predict specific years because he genuinely does not know, and he does not want to dilute the core message with a seemingly precise schedule.
The steering wheel metaphor
Those who oppose AI regulation argue that regulation would slow innovation and put the US at a disadvantage in the AI race, so the priority should be to move fast first.
Hinton's rebuttal: the metaphor itself is wrong. What they describe is not "driving faster," but "driving a car without a steering wheel, and accelerating."
Speed itself is not the problem. The problem is whether you can control the car.
"We still have a lot of control, and we should carefully build it in a way that allows us to continue to exist and coexist harmoniously with it," he said.
For the current direction of regulation, that is a fairly conservative expectation. The Trump administration earlier this year rejected most federal AI regulatory proposals, shifting regulatory authority from the federal to the state level, effectively weakening uniform safety standards.
"We don't know"
The conference theme was AI's social impact and international cooperation. One sentence from Hinton's speech mattered more than everything else:
"We don't know whether we can coexist with superintelligent AI. But we are building it."
"We don't know" — that is a Nobel Prize-level AI scientist speaking at an international conference. Not a skeptic, not an anti-AI activist, but one of the founders of the field.
He made it clear: we do not have sufficient reason to believe everything will be fine; we do not have enough tools to ensure safety; but we are moving forward at tremendous speed.
This is the part of the 2026 AI narrative that mainstream voices amplify the least. Silicon Valley's narrative is "AI will make everything better." The government's narrative is "we must win the race." Investors' narrative is "this is the biggest opportunity in history." Hinton's narrative is: first ask whether we really know what we are doing.
No one has a complete answer to that question.
Source: Unregulated AI like speeding with no steering wheel, AI godfather Hinton (Digital Journal, April 21, 2026, CocoLoop, AFP wire)