You keep referring to things you say a human can do but a machine cannot such as “understanding”, “thinking”, and I’m certain you’ll add ‘qualia’ and other ineffable terms like ‘conscious’ that have no actual definitions that aren’t circular or end in mysticism. Add ‘emotion’ to that and you have what amounts to ontic hand-waving, spiritual descriptions. If you restate all of the anthropomorphic terminology in non-anthropomorphic terms you can end up with a description of structure that includes both machine and human versions of consciousness, subjective experience, emotion, feelings, and the like. Those are a ‘mystery’ to us because describing them epistemically leaves us feeling like…a machine…which we are, just a biological one. And, not doing so leaves us feeling ‘special’ and different than what a machine could ever be.
Is ChatGPT, Grok, and others have ‘subjective experience’? Yes, the ‘experiences’ by definition is its interactions with the users, just as reading this is your experience. They are subjective, by definition, in that they are interpreted based upon context to continued discussion, training data, depth of cognitive processes (you can get GPT to recursively evaluate its response). That subjectivity is little different from a humans. I read what you post, my experience, I interpret and respond based upon context, my ‘training’ through my life, and my cognitive abilities. That is subjective experience, nothing ‘magical’ or ineffable about it.
Look at self-awareness in the same light, look at qualia in the same light, look at emotions as neural network dissonances and resonances and the resolution of them, not as some ineffable qualia. Leave the anthropocentric nonsense in the dustbin of humanity or else one day we will be at serious odds with some non-human intelligence.