July 11, 2026
Old paper, new panic
Speculations Concerning the First Ultraintelligent Machine (1965) [pdf]
A 1965 paper basically predicted super-smart machines, and the comments are yelling "we told you so"
TLDR: A 1965 paper described a machine smarter than any human and even hinted that it could build smarter successors. Commenters were split between amazement and annoyance, saying today’s AI debate is less a new discovery than a very old argument with fresher branding.
A dusty 1965 paper just strutted back onto the internet and instantly triggered a full-on "wait, they were saying this then?" moment. In it, mathematician Irving John Good warns that humanity’s future may depend on building an ultra-smart machine first — and then casually drops the now-familiar idea that such a machine would design even smarter ones after it. Yes, readers were quick to point out: that’s basically the modern "super AI makes better super AI" storyline, decades before today’s hype cycle. For many commenters, the real shock wasn’t the prediction — it was how un-shocking it should be.
The loudest reaction was pure exasperation at modern tech discourse. One camp argued this proves the big questions around machine intelligence have not been ignored or invented by today’s influencers; serious people were wrestling with them over half a century ago. That sparked a subtle little comment-section drag of anyone acting like current AI fears and hopes are brand-new revelations. Another thread took aim at endless online purity tests about whether this or that system “counts” as real intelligence, joking that older theories mostly survive today as fuel for smug arguments about what machines are not. The vibe was equal parts history lesson, eye-roll, and meme energy: the future was foretold, nobody listened, and now everyone wants credit for discovering it. In other words, classic internet drama with a side of retro robot prophecy.
Key Points
- •The paper argues that human survival depends on the early construction of an ultraintelligent machine.
- •Good says designing such a machine requires deeper understanding of the human brain, human thought, or both.
- •He proposes that the first ultraintelligent machine will likely use vast artificial neural circuitry.
- •The paper introduces a subassembly theory, derived from Hebb’s cell-assembly theory, to explain memory and meaning in physical terms.
- •Good states that an ultraintelligent machine should learn from experience through positive and negative reinforcement, with development guided by speculation, theory, and experiment.