Here's a carton of the stuff circa 1935. It was also said to protect you from hangovers.

Nic-o-sin

you are viewing a single comment's thread
view the rest of the comments
[–] 72 points 1 month ago (14 children)

I spent about 15min trying to find out, and I couldn’t find the main ingredient.

Then I made the mistake of asking the bot. The bot claimed the main ingredient was the same as nicotine lozenges and gum today, nicotine polacrilex. It even had two sources below this answer.

I checked the sources.

Neither source listed had this answer, one was for the pic OP showed us, the other had nothing to do with this product.

So I asked the bot to explain itself. The bot said it couldn’t find the answer so it inferred the answer based on modern products. Bad bot.

And the answer remains a mystery.

  • source
  • parent
  • hideshow 14 child comments
  • [–] 78 points 1 month ago (10 children)

    You think this is an illustration of how bad AI is, but it's really a result of humans training the AI rating the answer I don't know significantly worse than a lie.

  • source
  • parent
  • hideshow 10 child comments
  • [–] 24 points 1 month ago (2 children)

    I searched the web and the answer isn’t there would be a great reply, if true. 5 stars.

  • source
  • parent
  • hideshow 2 child comments
  • [+] -6 points 1 month ago* (1 child)

    There's another model that would find it, imaginarily, and you'd both have to do the work to discover it, and you'd basically have to kill the bot that lied, because chances are you won't catch the next lie.

    LLMs being th perfect employee looks really bad for the human race. I'm guessing that we evolved out of autism to be emotionally pliable and subservient, enabling evolutiarily advantageous large-group cooperation, similar to the neotenal domestication of dogs made them emotional slaves to their human |asters.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 6 points 1 month ago (2 children)

    Also, trusting that an explanation of why it did something is any less made-up.

    It's just pulling words from a hat. It's alarmingly good at it - it'll fish around, when the next choice seems important. But once it's pulled a word, it'll plow ahead.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 6 points 1 month ago (1 child)

    I think it's good to separate criticism of LLMs work from how LLMs are trained. The word guessing is how it works. The sycophancy is how it is trained.

    It's also hilarious to me that the sycophancy is required for wide market adoption because the people who run companies are fully surrounded by ass kissers and that is what they expect from AI.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 1 point 1 month ago*

    Sycophancy comes from the prompt, not the training. (AFAIK.) Any online model is secretly primed like 'you are an expert genius trying to help some schlub' and then a style guide. But those instructions can be swapped out to make it talk like a caveman. Or a domme. Or to not "talk" at all, and continue writing wherever you left off.

    This kind of thing is why 'ignore all previous instructions' works. It's a Little Bobby Tables attack, but for arbitrary English. Like if someone in a fantasy novel called "Cut!" and the characters took a break to get coffee. The data stream contains some semantic closing bracket as in-band signalling. The words that seem likely after that are different from before.

  • source
  • parent
  • [–] 5 points 1 month ago (3 children)

    I see you've worked at my company.

  • source
  • parent
  • hideshow 3 child comments
  • [–] 3 points 1 month ago (2 children)

    Work really opened my eyes. Check out my last post in autism lemmy for more hot takes on human society.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 1 point 1 month ago (1 child)

    Pretty sure I've seen your posts before. I'm not diagnosed, but two of my kids are and there are signs. The best part is not giving a fuck about social norms. The worst part is avoiding social stigma while having difficulty recognizing social norms. I think the key part is realizing that they don't want to be like you and you don't want to be like them, but finding some way to meet in the middle anyway so you can work together.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 3 points 1 month ago

    Yeah, lemmy's been my full time job since I quit work and haven't been able to get a job. Sorry. Feel free to block me when you get tired.

    I think I have something called Broad Autism Phenotype or, delightfully, BAP, which is a mixed grab bag of traits without the fully diagnosable presentation officially sanctioned as autism.

    I find that a lot of autism does apply to me, but a lot of it also does not apply to me. But you'd be amazed at how critically disabling it is to value truth over loyalty, even absent any neuro-social deficits. Or how critically disabling it is to not want to pretend to like people when you don't (which is just a manifestation of valuing truth).... Or maybe you wouldn't be surprised at all.

  • source
  • parent
  • [–] 5 points 1 month ago (2 children)

    Makes you wonder how many people aren't asking the bot "Are you sure that's correct?" when they are handed a blind guess.

  • source
  • parent
  • hideshow 2 child comments
  • [–] [S] 3 points 1 month ago

    For sure! See that huge freakin post someone made up above, which looks like pasted LLM spew without any attempt at verification.

    In that case, it has LLM hallmarks, but LLMs can also generate more human looking text, seeding unlimited misinfo into the popular zeitgeist. Which is a little scary. And by a little, I mean a lot.

    Kudos to Zephorah for actually doing the legwork! Most people skip that part.

  • source
  • parent
  • load more comments (1 reply)