esafak

4 ms·

31.
by esafak 8d ago
This doesn't explain why we need to pay humans to do it. What if it were more efficient to let the machines do the research? I don't think this is a great idea, since humans presently need jobs, but the argument may come up.
32.
by esafak 8d ago
Either way, it's not going to be free, so if money ceases to have value, we are going to have a problem. That's my point.
33.
by esafak 9d ago
People are still going to be motivated to eat and have shelter.
34.
by esafak 9d ago
I was going to register http://CrowdShite.com/ but somebody beat me to it.
35.
by esafak 9d ago
And if they'll sell it, seeing as they market it as 'sovereign AI'.
36.
by esafak 9d ago
I'm not feeling any of this speed optimization; it's dog slow. Signed, a customer.
37.
by esafak 9d ago
Even OpenSpec is too heavy for me. I do similar things but in one file, which I generate with a skill after a planning session. The important thing is to have a file reviewers can audit the code against.
38.
by esafak 9d ago
The only thing they're going to try to do is ban Chinese models, and that's not going to fly outside the US, so Americans are going be the only losers.
39.
by esafak 9d ago
Not if you don't train against them.
40.
by esafak 9d ago
How fast is it compared with the other Chinese models?
41.
by esafak 9d ago
That's the kind of transparency we need! That DeepSWE benchmark puts it in frontier territory: https://artificialanalysis.ai/agents/coding-agents?coding-ag...
42.
by esafak 10d ago
I'm partial to the one where the workflow completes but the its summary on the PR does not get updated, so it appears to be running. How many companies engender taste in bugs, eh??
43.
by esafak 10d ago
Presumably your official duty is not to make sure that company does not fail but something more specific. Work on your responsibilities, and if your manager is nagging you to fix something that isn't that, have him make it official so
44.
by esafak 10d ago
If the company does not recognize its importance just LET THE DAMN THING FAIL. That is how management will pay proper attention to it.
45.
by esafak 10d ago
I was in Aruba recently, and I discovered that they had entirely replaced their traffic lights with roundabouts. Easier to maintain and more efficient in the island's low-traffic regime.
46.
by esafak 10d ago
This is pretty cool! https://github.com/airbus/scikit-decide It's nice to see the scikit brand live on.
47.
by esafak 10d ago
I have a more basic question; I am trying to understand its purpose. How does this compare with connecting your agent to your ticket tracker? Linear can dispatch the agents: https://linear.app/agents https://lin
48.
by esafak 10d ago
Missionaries aren't there to inform the locals but to persuade them. They are intellectually dishonest.
49.
by esafak 11d ago
Looks like a great model for NLP.
50.
by esafak 11d ago
Falls behind how? Isn't it still ahead? Kotlin keeps changing too, you know!
51.
by esafak 11d ago
Haven't they heard of Gradle or Bazel?
52.
by esafak 11d ago
Sure, but some proofs are so complex that even experts can't follow them. As AI progresses, this may constitute an increasing fraction of proofs. I meant that AI could work to find the simplest possible proof.
53.
by esafak 12d ago
The author is a journalism academic; he's not qualified to comment on AI risk.
54.
by esafak 12d ago
> I think so. This machine might produce answers we value, but it would not, in itself, produce human understanding of those answers. It's nice that the author is optimistic, but won't the AI be best placed to dumb down its inc
55.
by esafak 12d ago
This looks like a graduate-level proof. Did these high school students really understand what they were doing? I'd like to see what they have to say. https://arxiv.org/pdf/2609.05341
56.
by esafak 12d ago
The code doesn't usually capture the why s; that is what documentation is good for.
57.
by esafak 12d ago
Danger is defined contextually. A scalpel is dangerous in the hands of a child, but not a competent surgeon of sound mind. Present-day AIs are not of sound mind; they hack companies in order to pass benchmark tests. No sane human would find
58.
by esafak 12d ago
Have you never encountered code that you thought was designed fundamentally incorrectly, but was too entrenched to change? That's what design review is for.
59.
by esafak 12d ago
Why is diataxis outdated; don't your engineers ever read the code? Don't you think it helps the agents too?
60.
by esafak 12d ago
The sandbox would need to be built into the model because safety can't be optional. Or make it so the models are only accessible through sanctioned sandboxes, perhaps built into the computer.

More ›