21 pointsby binyu3 hours ago9 comments
  • ido3 hours ago
    Any subject matter experts care to chime in with your gut feeling about where this sits in the "promising potential treatment" to "AI psychosis" spectrum?
    • Avicebron2 hours ago
      It's closer right now to interesting hobby project from a guy who's trained in chemistry, like a guy who's a mechanic keeping a project car.

      Maybe he publishes and it goes somewhere, I have a feeling getting approval for trials will be challenging with his setup as it is now.

      I do like the "do chemistry the old fashion way" ethos though. I wish he added a blurb about why he's doing it this way

      • cucumber37328422 hours ago
        >Maybe he publishes and it goes somewhere, I have a feeling getting approval for trials will be challenging with his setup as it is now.

        Yeah, drug development is one of those things where practically the only way to make it to production is to let a big incumbent with a big enough war chest to drag it across the finish line buy it.

        • skeledrew2 hours ago
          If it isn't a potentially big money-maker said big incumbent won't touch it though. Will more likely try to shut it down especially if it overlaps with any of their more ludicrous offerings. And given the properties that this has (easy to manufacture, stable, soluble, low/no toxicity, etc) it's ripe for a shutdown because it'd make AD treatments far too cheap.
          • CoastalCoderan hour ago
            Any reason he couldn't try advancing it somewhere like China or India, which (I assume) are less vulnerable to that kind of interference?
            • 6274679 minutes ago
              Big Pharma dynamics don't exist everywhere?
    • rtkwe2 hours ago
      "Potentially promising but likely to fail to replicate in human/non-mouse trials" is the easy default for new drug announcements, it's the modal outcome for drugs in this level of development, we've cured mouse cancers dozens of times over the years after all. The guy seems to have relevant expertise in the field so less likely to be purely AI driven nonsense.
  • amelius2 hours ago
    What does the guy's basement have to do with it? We live in the internet age, we can do anything anywhere. Was there not any other more relevant information to fluff up the headline? By the way, why are we still using twitter?
    • thunky2 minutes ago
      > What does the guy's basement have to do with it?

      Nothing, because per the post he actually did it in his garage.

    • 2 hours ago
      undefined
  • jorisw2 hours ago
  • WarmWash2 hours ago
    Painfully editorialized headline.

    He used AI to program robot arms.

    • throw3108222 hours ago
      "Meanwhile, LLMs (mainly ChatGPT Pro) were deeply integrated into virtually every step of the discovery process - I wouldn’t have been able to complete this project without them"
  • comboy2 hours ago
    Guy says he created a drug to treat Alzheimer's disease. Huge difference.
    • cristyansv2 hours ago
      > All of the in vitro screening was performed by an OpenTrons OT-2 liquid-handling robot programmed by Claude Code. Meanwhile, LLMs (mainly ChatGPT Pro) were deeply integrated into virtually every step of the discovery process - I wouldn’t have been able to complete this project without them
  • CPLX3 hours ago
    That does not seem to be an accurate description of what has happened here.

    What it looks like is someone with significant biochemical experience and a Harvard PhD has created some kind of drug or chemical that he thinks will be effective for treating Alzheimer's, and that he mentions using Claude Code to help him program some of the complex chemical engineering machines that he used along the way.

    • microgpt2 hours ago
      Shush, you'll pop the bubble
  • mikelitoris3 hours ago
    How did LLMs not immediately shut down for “biosensitive” prompt or whatever bs they build into them nowadays?
    • binyu2 hours ago
      That behavior was only observed with the short lived Fable preview, Opus and other models do not seem to pose such limitation.
      • hhh2 hours ago
        They still have biology safeguards.
        • binyu2 hours ago
          Can you provide an example or a reproducible way to trigger it?
          • mikelitoris2 hours ago
            I feel like if you asked it how to make nerve gas or something that would trigger it. But don’t do that.
    • DonsDiscountGas2 hours ago
      Opus doesn't do that (at least it never has to me). I'm pretty sure codex doesn't either. It's just Fable
    • microgpt2 hours ago
      Because he only programmed robots to move A to B and didn't actually do any biology with AI
  • jqpabc1232 hours ago
    With help of AI, i developed a universal cure for all disease --- strychnine.

    However, it does seem to have a few undesirable side effects.

  • negergreger2 hours ago
    [dead]