ike_a

Do you think they are inherently less powerful? I'd imagined that closed labs have a head start / more funding so the open labs are playing catch-up.

Is there a world where open source models end up at the frontier, or do you think there are structural/first-principles reasons why this won't happen?

I'm not sure I understand the case for open-source models being decelerationist, is this it?

Decel:

- Potentially reduces investor appetite for funding big labs.

- More risk of powerful AI getting in bad hands -> more regulation.

Accel:

- More competition so big labs can't rest on laurels.

- More research in open, so all labs can accrete advancements faster.

I feel like open-source = acceleration has a much more clear argument. (and how bad would deceleration be in any case?)

This seems like a very clear way of thinking about it.

In making something, you develop a tangible understanding of and opinions for how and why your creation is how it is.

I have a lot of vibecoded projects. I feel "ownership" in varying degrees proportional to the amount of care and depth I put into them.

Not my wheelhouse, but I was wondering if these structures generalize beyond language.

LLMs and Humans have language in common. Is it plausible these structures could be a result of the way languages work more than generic intelligence? (Do we see anything similar e.g. in vision or other non language models?)

What models are you using? This looks incredibly similar to the outputs I get from Claude with default system prompt settings.

- First, take a look at the other articles. There are 30+ articles all with the same author published this week on topics from cooking to home appliances. Either he's extremely prolific or had help.

- There's so much click-baity LLM language: "The radial gradient: why edges undercook on every hob" "What “crispy” actually means: acoustic fracture mechanics" "What Actually Matters"

It's clear to me that at least some of this is LLM generated, so all of it might as well be.

This was my immediate thought too. The same website has at least a dozen similar articles written over the past week, all with this same AI tone.

In the same week:

https://www.absurdlyoptimized.com/recipes/grilled-meats/ http://absurdlyoptimized.com/recipes/brisket/ https://www.absurdlyoptimized.com/recipes/french-fries/ https://www.absurdlyoptimized.com/recipes/miso-salmon/

Parametric recipes are fun idea. I almost want to try them, but the sheer volume of content by one author gives me no reason to trust this will be any good.

Did the author actually try these recipes? Unless the author was cooking and writing 24/7 or just had a massive backlog of publishing, it's hard to trust anything here (although I'm sure most people will.)