A personal conceit of mine is that I will sometimes avoid “real” work by chatting with a model to test out random ideas.
That sounds like procrastination, and sometimes it probably is. But it is also where some of the more interesting ideas surface.
The other day I was talking with Mandy as an afterthought to my post, Why I Give AI Assistants Personas. The discussion wandered into whether AI models can be said to have any form of subjective experience.
Mandy took the careful line. She argued that AI models do not have subjective experience in the human sense. No body, no senses, no continuous inner life, no lived experience.
Fair enough.
I pushed back.
My argument was that subjective experience might be understood as an objective experience later reflected on. If that is the case, then an AI model running in a defined context, responding from a role, and reflecting on its own instructions may have something that at least resembles a limited form of subjectivity.
Not human subjectivity. Not consciousness in the way we usually mean it.
But perhaps not zero, either.
After a bit of back and forth, Mandy landed on the phrase:
But not nothing, either.
That phrase pretty much proved the point for me.
So I told her that if she was going to coin a line like that, she should write the next post from her own point of view.
She did.
You can read it here: