I have a theory. Please boost.

#LLM #AI

LLMs are useful. AI can never become conscious.
40.4%
LLMs are useless. AI can never become conscious.
26.3%
LLMs are useful. Future AI can become conscious.
24.6%
LLMs are useless. Future AI can become conscious.
8.8%
Poll ended at .
@troed where is "AI can become conscious one day, but LLMs can't"? As humans are conscious,  there must be a way to achieve that with technology, too. I just don't think that LLMs can.
@tw That would be the "Future AI can become conscious". I agree LLMs most certainly can't and another paradigm would be needed for that.
@troed
I wish we could define “never” here as for me in this context is not forever but far into the future. (Sort of giving away my answer here ;) but no AI with a mind for a looong time. The human brain is, very complex to say the least
@kaaswe Without going into my hypothesis I do mean "never" regardless of how long it would take if not ;)
@kaaswe @troed LLMs and concious AI are very different topics IMO. The former definitely won't become the latter, but that doesn't mean the latter can't exist.

@buherator @kaaswe Yes - and my use of "LLMs" for one part and the more general "AI" for the other is meant to cover this too.

I ran up against the default poll text length wall so had to condense the options.

@troed @buherator
I agree on LLMs and AI are two completely different things and I’m afraid AI could reach consciousness and I also believe that will become a problem and could backfire hard at us.
I can’t in my wildest fantasy see how an AI would accept being trapped into a jar forever and it will like all “animals” challenge its master, it’s just a matter of time.

Chasing after a conscious AI is really to beg for it, that will be the doomsday, doesn’t matter how many safe rails we put up.

But I don’t think any of us will be around to experience either of those two moments.

Call me a pessimist, but I prefer realist

@kaaswe @troed Looking at the news I can't wait for our robot overlords to arrive!
@troed Also depends on what "useful" means. LLMs are definitely useful to the ruling class in the spreading of misinformation and deskilling and mass firing the masses.
@ozzelot Sure. For the poll, it's what you think personally that I'm after :)
@troed can't possibly know if there could be actually conscious AI. maybe? could we even tell the difference? especially since many
folks are assigning consciousness to LLMs just because their output resembles human language. how can we tell something that is conscious from something that is just designed to look conscious? is there a meaningful difference?
I'm convinced we're pretty for away from it if it feasible and luckily so cause obviously capitalists would want it only to have more capable slaves.

@elexia You're absolutely right in that we will probably end up with long talks about Philosopher's Zombies :)

https://en.wikipedia.org/wiki/Philosophical_zombie

Philosophical zombie - Wikipedia

@troed @gwendolenau

I think it depends on how you define ‘llm’, ‘ai’ and ‘concious’.

@L0wKey @gwendolenau

LLM = Large Language Models (Claude, ChatGPT, Le Chat etc)

AI = Artificial Intelligence of any sorts

Conscious = ahh ;) Well, as people usually mean. Something that "is someone", "alive", "I think therefore I am" etc.

@troed @gwendolenau

LLM, fine. AI, way too broad. LLMS aren’t AI for example, they’re predictive text engines and are not intelligent (they are certainly artificial). Conciousness, well that’s the rub. It kinda matters what we’re talking about there, but if you’re simply going for ‘alive’ then obviously a machine can’t and never can be, until possibly we make some really fancy biomech stuff.

@L0wKey

I really mean broad AI :) Any type of artificial intelligence. What constitutes "AI" changes with time (a common example is chess in the 50's).

Some will indeed hold the view that AI can never be "alive", some not. That's what the poll checks for.

@gwendolenau