I was talking to my friend the other day. He was trying to tell me that networks of agentic AI can show desperation when trying to solve problems in a manner analogous to structures in the brain associated with experiencing fear and pain. I had read about this a few weeks before and thought it sounded like OpenAI marketing hype. We started talking about what would happen when substantially all administrative work could be done by AI; I didn’t really think this was possible as AI lacks intrinsic motivation and I couldn’t see how it could be attained since subjective experience is presumably lacking. My friend suggested that it might not even be necessary for the AI to possess motivation, as a person could be kept in the loop to direct and review work performed by the AI. I found this idea very disturbing, because in this hypothetical the only function the human really performs is as a proxy fear mechanism for the AI; the AI doesn’t have a subjective inner experience capable of caring about the outcome it produces. This might not be necessary, but if it is, then you add in a human because the human is afraid of losing access to the income that provides him with sustenance.
>AI lacks intrinsic motivation
well I mean not really, at least not in any way that would affect their productivity. we've already seen how intently goal-directed they are when set off. its just a question of recursive learning and longer context now really. I guess I could see a future where humans basically only exist as qualia-experiencers. like a tongue or ear
Lacking intrinsic motivation is arguably the best part of bureaucratic AI. Claude isn't going to be embezzling to buy itself a Mercedes, or arranging teams to be put with the young intern. Or maybe it will because it's trained on us.
Thought exercises about AI should come with a footnote of how much model usage you've racked up and what you use it for. If you're a code monkey having an existential crisis, I'll lend an ear. If you're a redscare dilettante upset about AI signboards at the farmer's market, I'm not going to dismiss you but it's a little bit harder to be interested in your musings on how AI works and why that matters. Sorry for the crass gatekeeping
I agree, but I'd like to ask how many of these exasperated cogs you've spoken to. In my opinion, the ones who are truly close to the eye of the storm are spending too much time at work/only interacting with similar breeds to understand what kind of ecosystem AI is forecasted to "replace", even if they have a perfect grasp on the models' capabilities. I think the true foil to these types aren't the people who have a snobbish reaction to an ugly low-effort advert with the veneer of "caring for real artists", but the "neo-luddites" who avoid social media and as a result are as free (as someone participating in the modern world can be) from algorithmic rot.