In fairness the Google LLM integration is mediocre and I don't trust it. One reason I think we're at the low point for AI sentiment is that we're clearly at the height (I hope) of tasteless LLM integrations. People are flinging shit at the wall and hoping it sticks, in the meantime it just stinks.

Or maybe it might make more sense to say peoples feelings will change 10-20% and their justifications for the feelings will change 60-80%, if you get what I mean. :p

Oh sure, I'm more imagining something like an 80% reduction in reliance on this kind of argument and an uptick in different arguments that lead to much the same overall position. It's not that people's feelings will change overnight, just the rationalizations they use to justify them.

What might make sense is to write various vignettes exploring phenomena that happen around each sort of model. The ways in which big platforms exploit user ignorance, the dragon for an open platform dealing with a tricky bot network, the internal arguments around antiscrapers using AI to detect AI.

To most people whose intellectual background is some form of 20th century humanism they're basically a worldview shattering nightmare that doesn't offer much to replace it. Just kind of a blackpill. This will change, but right this minute they're very easy systems to hate with obvious flaws.

Though I do in fact think AI sentiment is probably at its low point. Right now the systems are both freaky because they make it clear that the sparks of human creativity really are mechanical but also fundamentally uncharismatic and unenlightening so they don't really make up for it with anything.

I didn't say they'd accept it, I said they would concede that it has utility and is not just a slop machine. This will only be because the point will be so obvious that to deny it would be delusional, relegated only to echo chambers and the most hardcore kind of anti-AI rage poster.

Because of the natural incentives around data access, it is generally understood that AI systems in this role are a kind of outgrowth and extension of the network and they have different agent strategies for running a platform much like the commerical, open, and private categories I outlined above.

This makes the selfish purpose of human contact for AI that humans have reliable access to the world outside the computer and AI agents do not. To get access to this knowledge it is useful for AI agents (or their human sponsors) to situate themselves as the curator of a human social network.

While these agent systems generate mountains of grounded long text and are therefore quite good at reasoning and programming tasks, they still have limited access to the outside world since robots are expensive and people are uncomfortable just letting them walk around doing stuff.