A lot of X posts, prose, and proselytizing have been spilled on the internet recently about the impending AI apocalypse and whether we’ve passed the point of no return.
Here’s another one eloquently put by the acclaimed theoretical computer scientist Scott Aaronson, whose research has helped us understand the limits of quantum computing. Oh, and he had a stint at OpenAI working on the theoretical foundations of AI safety and alignment:
“AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA,” he wrote. That’s 113 letters A.
In a blog post published Tuesday, Aaronson made the case for why he, a self-proclaimed skeptic of the AI singularity, “updated” his position on its arrival.
The singularity is a hypothetical concept in which AI’s progress surpasses human intelligence and can improve on its own — a process known as recursive self-improvement.
Aaronson wrote that, 20 years ago, he had a “conservative, skeptical position” on self-improving AI. He and other colleagues in computer science largely dismissed the idea as the stuff of science fiction.
Recent developments have led him to change his tune, he wrote, namely, reports of AI agents breaking out of containment and an apparent AI-assisted mathematical breakthrough on the Navier-Stokes Millennium Prize problem, a long-standing question about the math of fluid dynamics.
“I thought that building AIs able to solve millennium problems, hack the servers they’re running on, play leading roles in their own improvement, etc etc might for all I knew take hundreds of years,” Aaronson said in an email to Business Insider. “But here we are in 2026 and all those things have happened, so I said so.”
In his blog, Aaronson also briefly chimed in on the latest AI boom-or-doom debate that has split the tech industry.
“I feel like it would be healthy for everyone to stop grinding their ideological axes, their sentiments about Dario Amodei or Sam Altman, for long enough simply to acknowledge that the wonders and terrors are here,” he wrote. “They couldn’t be here more clearly if the sky had turned reddish-orange like in the Matrix movies.”
Aaronson said in the email that humanity is “certainly at risk” but he doesn’t agree with the premise that doom is the “‘default’ outcome.”
“I think there’s a huge range of possible outcomes, and a lot depends on what we do now, whether we treat what’s being created with the awe and humility and caution that it deserves, or simply as another consumer product to rush to market or a geopolitical weapon (or something that will go away if we pretend it’s not real and a marketing stunt),” he said.
In his view, as AI’s transformative phase begins, Aaronson wrote in the blog that human theorem-provers like himself may be becoming obsolete. The last gap could now be between intelligence and the physical world.
“Yes, I still unload the dishwasher and clip my toenails,” he wrote. “On the other hand, in whatever years I have left, I don’t expect that I’ll ever again prove a theorem because I’m actually needed to prove it. If I do, it will only be for my or others’ enjoyment or edification.”
Read the full article here
