[ExI] The biggest advance in Nanotechnology in decades

[Ben Zaiboc]

On 31/05/2026 13:00, John Clark wrote:
> For years a big question in my mind was which would come first, AI or Nanotechnology, about four years ago I discovered that the answer was AI. The problem with Nanotechnology was that although scanning tunneling microscopes were able to give you complete control over atomic placement of atoms, they could not simultaneously give you complete control over atomic chemical binding, and many thought they never would because of the so-called "Sticky Fingers" problem. But last Thursday a paper was published that I thought proved that the Nanotechnology skeptics were wrong


Hm, I don't think that this shows you were wrong (and wasn't the 'sticky fingers' problem disposed of years ago anyway? I thought it was a non-issue now). It's an advance, certainly, but it's still not delivering nanotech (molecular manufacturing).

AI will probably be needed in order to actually realise nanotech (especially considering the complexities of actually controlling nanotech. systems).

I think that our efforts would be best put into developing AI first. Once we have it, then with the kind of thing we talked about previously (large-scale robotic biochemical experiments. which can be extrapolated into large-scale robotic experiments of any kind), a feedback loop can be started with AI driving advances in physical experimentation, driving advances in computing technology, enabling more powerful AI, etc.

LLMs might be a start, or they might be a massive red herring, but I don't think there's any doubt that we need actual AI, as quickly as possible.
-- 
Ben