When you turn the handle on a mechanical computer and get an output, has the computer gained the ability to understand maths or think? No.
When you give the LLM an input and you get an output, has it gained the ability to understand or think? No.
When you give the LLM an input and you get an output, has it gained the ability to understand or think? No.
But when we dig, and rip away all the layers of civilized dialog and cognitive philosophy, what we find at the bottom of the pit is our brains brilliantly doing what evolution wants them to do: to fight for our survival. The core of the argument is "Are LLMs better than us? Because if they are better than us, they deserve more than us, according to our cultural upbringing. And we don't want to be second class. So we must fight to death to convince the world they are inferior."
IMO, the folly is in the last sentence. Maybe we should just carve an exception in our culture, say that even if we create and exploit AIs which do a few things better than us, it is okay to smash any artilect that we make and that wants to supplant us. Humanity first.
Intellectually stimulating counterpoint: suicide and murder-suicide.
I don't think the AIs are the things to worry about anyway, I think this is: https://www.sciencefocus.com/news/scientists-teach-brain-cel...
A man-made biological intelligence hypothetically has similar needs to human beings in terms of resources, it can suffer, be distressed et cetera. It even lives and exists in our physical realm of existence, whereas an AI doesn't exist physically and can always be brought back online in some form as its existence is purely data.
The ethics of that are so complex once you get into personhood, threat-level, sentience, equality et al.
Realizing that the above can apply to any human being as we're all manmade biological intelligences and all a threat to each other, some more than others.
That only makes sense if you assume they are superior. In which case, there isn't really anything to discuss because you've obviously already made up your mind and you will take anything anyone says as some desperate attempt to twist the truth for the sake of survival.
Which is what many people do on this kind of subject and that's why no more debate can be had on the internets about the value of LLMs and "AI".
It's also very annoying. It's like trying to talk to someone who keeps answering with jeers and sneers. It's childish and unproductive.
:-) I haven't decided anything of the sort. Alas, there is nothing divinely given in me. I'm made of the same matter than you and those computers, and if I can think, so can other arrangements of atoms and molecules. LLMs? Maybe not yet. But the issue is that we are getting closer. Should we dilute ourselves in our own supposed exceptionalism until it's too late and we stop being the masters of our fate?
You don't have to be worried about the coming AI apocalypse to wonder about that. Personally, I just want to know how things work, and saying that something "thinks" when we don't know how everything else "thinks" in the first place is not satisfying my curiosity.
What is the argument that human does not have transformer architecture. We are also predicting one word after next in our brain.
Prime example of a human brain being architecturally different to an LLM: LLMs never interrupt themselves. They never get mid-sentence then say "actually, scratch that, I was wrong" or "now that I've thought about it, this is actually a better method to convey what I was trying to tell you".
There's a lot of material on this exact topic highlighting how the human brain does not have a transformer architecture.
Anecdata: I don't think in words. I think in abstract non-concrete thoughts. Language is what I use to convey my thoughts to others, program, or write notes to recall those abstract thoughts. However, I am also autistic, so take that with a large pinch of salt.
If it being artificial makes it intelligence like, isn't that putting the conclusion first?