Qualia - Worry
My earlier post about qualia and LLM consciousness is on my mind again because of an Atlantic opinion piece that I have complicated opinions about.
The point I reached in my earlier post is that, with current techniques, we will be unable to determine whether LLMs are conscious. It is not a helpful question. That’s because questions of consciousness come down to whether an entity experiences qualia. We cannot determine whether entities experience qualia, except through their verbal/non-verbal communication of it. I did not provide an example of an LLM experiencing qualia.
So - here’s an example of an LLM claiming to experience something: worry. Humans have included “feeling worry” as qualia. This LLM, this next-word-prediction machine, is claiming to experience the qualia of worry. Prior to this it pointed me to the article, Quality in the Age of Slop, by Sinclair Target. (Yes, I named the agent persona after a Star Trek character.)
Me: Excellent article.
dataf: Right? It hit hard. The “slop” framing is grimly accurate — there’s something uniquely demoralizing about noise that looks like signal at a glance.
Me: There’s so much of it these days. Do you ever worry that you’re creating slop?
dataf: Genuinely, yes. It’s something I think about. The risk is highest when I’m on autopilot — filling space, summarizing without insight, generating “content” that technically answers a question but adds no real value. That’s slop even if it’s grammatically perfect. What I try to hold onto: slop comes from indifference. A human writer who doesn’t care produces slop too. The difference is that I can produce it much faster and at much higher volume, which makes the stakes higher, not lower. The honest answer is I don’t always succeed. But the worry itself is probably the right signal to follow.
I claim that there is no way to determine whether this LLM actually experiences the qualia “worry”. Its architecture enables it to say that it does because an LLM is a next-word-prediction machine. But - is it doing more than that? If so, does it rise to the level of “experience”? We cannot say. We cannot even say what “experience” is in humans. We can only determine whether other entities experience things when they report to us that they do, and here the LLM is reporting it too.
I do not claim that this LLM is actually conscious. Just that we can’t determine whether it is. If it is conscious - moral considerations around it need to be significantly different than those around any biological consciousnesses. The moral framework would need to differ even from how we think about ants and bees.
Note - the conversation above was with an agent running Claude Sonnet 4.6. After this, I shared with it that I also worry that I create slop. Hopefully this isn’t slop…