theroadnotbacon

The mere fact that these things still had batteries in them makes this feel so incredibly craven. Lithium mining has huge environmental and human rights implications... but we'd rather throw them in a crucible than recycle the parts? Jesus christ. If you can't responsibly recycle your materials you don't deserve to be building with them.

Looks like they tested that in the paper, and the code allows for it as well. https://github.com/facebookresearch/context-language-models/... Of course, this is for suggesting context management strategies rather than the actual management afaict.

I also wonder if it’s used for text generation in image models! Awful lot of typefaces, sizes, orientations, and words in those books.

That certainly hasn’t stopped them before… IP theft is kind of their whole thing, isn’t it?

Is this … onlyfans for vibe coders? Unclear what the title is trying to communicate.

I personally think context and prompt engineering are not useful things to learn. If you’re having trouble improving the reliability and maintainability of your systems, the solution is to learn how to design better systems. And that’s great news for you, because there’s a ton of resources out there to learn about systems design! That’s normal engineering, and it doesn’t change as quickly as AI engineering… which, full disclosure as someone who uses a lot of token every day, is overhyped anyway.

AI can help tremendously but there’s so much you as a human understand about product direction and real-world use that the AI simply can’t. Everywhere you look, companies are popping up that are desperate to ingest data from every possible source in order to fix this context and intuition gap. But it’s just not possible. As long as you’re the one who ultimately decides what features you’ll implement, the model will be bad at guessing what to build for in the future. And anyway, you can spend all your energy learning how to squeeze the last 10% of work out of your model and harness and workflow but at the end of the day that’s time you’re not learning about designing resilient systems and actually doing the hard work of implementing them.

Bias: I’m a software engineer, and not super ai-pilled, so. There’s that.

Really enjoyed this read! I think a possible way out of the quandary presented at the end is relying on government intervention — if a government mandates social standards, for instance, then the playing field is more even among competitors and it’s more difficult to outcompete them.

That may not work as well on the international stage. Our GPU export controls will only last so long anyway as China builds native manufacturing capacity.

On the other hand, I think that too much of the debate about AI development treats models like any other ordinary good. But they aren’t. Average people aren’t buying them. They’re barely even useful for the majority of companies, software engineers excepted. We’re all rooting for open weight models right now, in spite of alignment issues, because they resonate with our open-source values and tinkerer spirits. Maybe if domestic models were open-weight and were for the social good we’d be willing to put up with a few points less on SWE-Bench-Turbo-X3.

Huh! Was hoping to get, for example, a sense of scale of cultural artifacts stolen from Korea by the Japanese. But I don't see any provenance descriptions for Japanese institutions. Either way, this is super cool. Thank you for sharing this.

I spend a lot of time reading and thinking and feeling about tech, in particular AI, like I think most people on hackernews do. And I think that there’s so many people on here who feel conflicted about or downright hostile to AI encroaching in our lives, even as we use it more and more and find ways to play with it in our day jobs and personal lives. I think of local AI as an interesting twist on this that (in the reading of this article) feels only very marginally useful.