mchinen

Ah, I mean the AI companies converged to look more similar to each other, and at the same time diverged from the original vision for safety.

Aren't the route restrictions that this data was collected on a large confounder? They are specifically placed to avoid difficult or dangerous areas. For example in the bay area only recently were they allowed to go on the freeway, and only in certain areas.

That huge impromptu 2023 party at the Exploratorium Hugging Face did had such amazing covid-unthawing AI spring energy, and I keep thinking about it. It was clear then they had the momentum to do a lot of things, and more so now. So I was surprised by this news.

The party is a is a relatively small concern compared to the openness and github comparisons, but since no one else mentioned it, I hope something like it happens again. Or maybe that moment has passed.

Today I noticed they also wall off the normal (non-old) site after the session goes on for a little while.

Hopefully it's just another A/B they drop after a week like in the past, but otherwise I may have to move on to another semi-anonymous doomscroller.

The protein folding solutions like ESM/Alphafold were not due to this LLM/agentic coding or autoresearch type approaches though. They were designed by bio ML researchers.

It's hard to keep track of the frontier on bio ML, but it seems that we're going slower than what Demis Hassabis said in 2024 with 5 years to full cell molecular simulation. We still aren't able to reliably model a tiny surface of the cell membrane.

And of course there's Derek Lowe's takes on the drug discovery pipeline waiting for the proof in the pudding.

To me the only reasonable bullish position is that there is a very non-linear AGI threshold for accelerating progress that we haven't hit yet.

For me personally, I'm looking at other more tractable fields as a proxy to measure this kind of progress. The best modest evidence is from the agentic coding area, (modest because these kinds of gains may not translate to bio progress). Other soft-ish fields to like legal/law/tax are also interesting to watch, as a small amount of people are now trusting AI for these areas that were considered totally unusable a year ago. Another proxy is being able to generate generally entertaining media.

SRE is a 'say no to powerful people' job as well. I think for this to work the leadership needs to show support for it. The friction is still there, but in more tolerable areas. I bet for ethics this isn't the case at OAI, but everyone values security and stability.

For an SRE there can be more directed hate received from the junior employees, that want to release new features they developed. Especially because there is less accountability across orgs. Security is an interesting one because it seems to have less of this friction, maybe because it's more clear cut what is an issue.

I've been considering different ways to measure code understanding and fluency. I'm still using traditional 'easy' problems on a coding pad to probe this, but I'm finding a new generation candidates that don't prepare for this, and may be able to have a full career with just a higher level of code understanding. But they might not ever develop the ride-a-bike with one hand level of fluency unless they grind pointless leetcodes.

So I'm increasingly uncertain about what and how to test. My default for now is still to rely on ability to write basic code fluently, but I'm open to changing this perspective.

I really want to know how this existing repo AI-assisted live coding test works, with example problems.

It seems the standard data structure puzzle type thing won't be feasible if you are using an LLM.

Also the latency for these agentic coding/prompts seems like it would make the interview a bit awkward.

Anyone been conducting or taking interviews with this kind of thing with thoughts to share?

I'd love tips for maintaining a good meetup.

I tend to like the smaller meetups that are focused. But they often degrade in quality as they scale. A failure mode I hit several times was group circles with a louder or opinionated person eventually entering and taking the focus of the group. For large meetups that's not an issue I suppose. Break-out groups did work, but it needs some planning.