> which could house qualia.
I postulate this is a self-negating argument, though.
I'm not suggesting that LLMs think, feel or anything else of the sort, but these arguments are not convincing. If I only had the transcript and knew nothing about who wiped the drive, would I be able to tell it was an entity without qualia? Does it even matter? I further postulate these are not obvious questions.
Transformers attend to different parts of their input based on the input itself. Currently, if you want to tell an LLM it is sad, potentially altering future token prediction and labeling this as "feelings" which change how the model interprets and acts on the world, you have to tell the model that it is sad or provide an input whose token set activates "sad" circuits which color the model's predictive process.
You make the distribution flow such that it predicts "sad" tokens, but every bit of information affecting that flow is contained in the input prompt. This is exceedingly different from how, say, mammals process emotion. We form new memories and brain structures which constantly alter our running processes and color our perception.
It's easy to draw certain individual parallels to these two processes, but holistically they are different processes with different effects.
When really they are an actual feedback mechanism, that can totally be quantified just like any control loop. This whole 'unknowable qualia' argument is bunk.
What's the difference between qualia and a soul?
Qualia can be an entirely physical phenomenon and is not loaded with theological baggage.
As an independent organism, my system is a culmination of a great deal many different kinds of kins, which can usually be broken down into simple rules, such as the activation potential of a neuron in my brain being a straight-forward non-linear response to the amount of voltage it is receiving from other neurons, as well as non-kins, such as a protein "walking" across a cell, a.k.a continuously "falling" into the lowest energy state. Thus I do not gain any conscious perception from such proteins, but I do gain it from the total network effect of all my brain's neuronal structures making simple calculations based on sensory input.
I don't know if LLM's will. But there are lots of AI models, and when someone puts them on a continuous learning loop with goals, will be hard to argue they aren't experiencing something.
But, it is still just a way to try and describe this process of processing the inputs from the world.
It isn't metaphysical, because it can be measured.
I might have said 'unknowable' a little flippantly.
I just meant, in these arguments, some people start using 'qualia' to actually mean some extreme things like our mind creates the universe or something.
It's one of those words that isn't defined well.
Can someone who's never seen red hallucinate something and assume it to be red? What if that red is correctly the red they would see if they saw red?
Can you reproduce this feeling in someone by doing something to their physical body without showing them red?
If so, how does it differ from the latent encoding for uploading an all red pdf to your favorite multi modal model?
Instead of doing that socratic bs you see a lot here, I'll be more direct:
Until there's some useful lines that can be drawn to predict things, I won't accept using a fuzzy concept to make statements about classification as it's an ever shifting goalpost.
There are answers to my legitimate above questions that would make me consider qualia useful, but when I first learned about them, they seemed fuzzy to the point of being empirically not useful. It seems like a secular attempt at a soul.
Now, obviously if you're trying to describe something with experience, it needs some actual memory and processing sensory input. Current Generative AI doesnt have a continuity of experience that would imply whatever qualia could mean, but I find it hard to definitely say that their encodings for image related stuff isn't qualia if we don't have hard lines for what qualia are
You can then get a distribution of what people think is 'hot' versus 'cold'. What is icy, versus, bearable.
When you go to a doctors office and they ask you on a scale to rate pain, do you think that is completely bogus?
It isn't exact, but you can correlate between people. Yes, red heads feel more pain, there are outliers.
But a far cry from metaphysical.
The problem here is the word 'qualia'. Its just too fuzzy a term.