Post by @mattsheffield@mastodon.social

2026-10-07 09:05 UTC

And even if the "models have a pain representation" argument were correct, having the ability to represent pain is not the same thing as a physical experience of pain which literally alters the physical bodies of biological entities. Chatbot internalities don't exist across answers. There is not evidence that chatbots have feelings or permanent selfhoods, but because they can write so well, this makes some people be easily misled about them. https://plus.flux.community/p/large-language-models-and-the-textual

Replies (3)

@NatureMC@mastodon.online 2026-10-07 09:27

@mattsheffield@mastodon.social A very interesting article about the #stochasticParrots. I bookmarked it for more reading. Just a note: I nearly didn't click it because I don't read articles with genAI pictures anymore. Fortunately it has real human material, too!

Open ##4949290
@sysop408@sfba.social 2026-10-07 13:12

@mattsheffield@mastodon.social the idea that Eliza could be taken seriously by anyone was immediately ludicrous to me from the first time I ran the program as a teenager. I kept hearing that people would actually feel like they were understood by the program so one day I decided to suspend disbelief and pretend I didn't know better. I ran the program and spent 30 minutes talking to it like I thought it was just an awkward person. After about 15 minutes, I started to understand how people could feel like this was a real experience. The further I got into the exercise, the more inclined I was to meet it half the way. People regularly respond incoherently and instead of mocking them, we restate what we meant. I started to do that and the conversation actually flowed. Of course it was mostly just me talking to myself, but it only took one or two lucky responses by Eliza to trigger a surprise hit of dopamine to make it feel more rewarding than just writing a journal. #AI

Open ##4950523
@HistoPol@mastodon.social 2026-10-07 14:08

@mattsheffield@mastodon.social "There is not evidence that chatbots have feelings or permanent selfhoods, but because they can write so well, this makes some people be easily misled about them." I have started to take a contrarian view: "... 👉#ArtificialGeneralIntelligence has already happened👈 (Blaise Agüera y Arcas, VP and CTO, #Google 1)) 👉#LLMsAreIntelligent! 👈(#GeoffreyHinton, #GodfatherOfAI 2))..." (Source/Thread:) https://mastodon.social/@HistoPol/116716598209024407

Open ##4951129