The difference here is that a cheap AI chip won’t fix the fundamental software problems with LLMs. We might reach a point where they can produce output faster, but as long as what’s actually going on is probabilistic next-token prediction in a static vector database, that just means faster mistakes as well.

There’s an odd psychosis going around where people become convinced that actual AGI can be derived from this technology. People who should know better just shut their brains off when it comes to token prediction, because they’ve had very compelling “conversations” with the predictor. They forget that the actual model is static, has no internal state, and doesn’t even “remember” what you’ve said to it.

What it has is a context window, and your entire conversational history - both what you’ve said and how it has responded - gets shoved into that window when you interact with it. (Or depending on the chatbot harness, saved in “memory” files that it can retrieve when the context contents indicate that would be useful.)

That’s why the bots seem so weirdly forgetful one moment and like they’ve got photographic memories the next. Stuff that is in the context window and has its “attention” will influence the tokens it produces, but whether or not the right things are in the context window and it’s including them in the token prediction is a crapshoot.

No, I’ve decided not to use a tool that isn’t fit for purpose. Studies have consistently shown that people think they’re much more productive on LLMs than they actually are, and the side effects of cognitive debt and skill loss aren’t worth the often imaginary productivity gains.

Combine that with the awful financials of OpenAI, Anthropic, SpaceX, and Google’s Gemini, and you’ve got the makings of a really nasty situation.

Say I come to depend on Claude Code for all my coding, and build an entire workflow around it. I get used to spending thousands of dollars a month for access, and tell myself it’s worth it for the imaginary productivity gains (although what’s actually happening is that I’m just producing a shit-ton of garbage code no one has any hope of understanding).

Then Anthropic reaches the point where they run out of investor money to spend, and fold because even charging developers like me thousands of dollars doesn’t even begin to cover their costs, and I’m one of their rare customers who are loyal, versus the bulk of developers who use up their free allotments of tokens and then model-hop.

Now what? My workflow is broken, I’ve forgotten how to code, and nothing I’ve produced recently is human-readable.

No thanks. I’m not an old man yelling at clouds, I’m a software developer who can recognize a problem when I see one.

If it does get better, it will be with technology other than LLMs, because LLMs don’t get cheaper per unit of usage as usage scales.

I suspect that we won’t have actually useful AI of some sort until LLMs get out of the way. They’re sucking all the oxygen out of the room right now.

What we want is software that behaves predictably. Since LLMs don’t do that, we don’t want them or their “agents.”

For fuck’s sake.

I had someone ask me recently if a picture of Thomas Aquinas was AI-generated. It was a 15th-century painting.

We’re sure they’re absolutely PR stunts and bullshit.

An LLM does nothing without being prompted. An LLM only has access to the tools you give it via whatever harness you’re interacting with it through. An LLM in an actual sandbox has zero chance to hack anything, especially if it’s properly air-gapped, as any responsible person would do with technology they actually think is dangerously powerful.

If their LLMs are behaving badly, that’s because their prompts are poorly written, their harnesses are vibe-coded garbage, and their “sandboxes” aren’t real sandboxes.

Hey now! I can honestly say Microsoft has never done that with any of my computers!

(Why yes, they do all run Linux. How did you know?)

AI is going to happen.

Oh? So there are hard numbers showing a positive ROI now?