scellus

7 ms·

1.
by scellus 8mo ago
the main bottleneck for the civilization in communications currently is the sparsity of cynical, negative HN comments
2.
by scellus 9mo ago
Perfect economic substitution in coding doesn't happen for a long time. Meanwhile, AI appears as an amplifier to the human and vice versa. That the work will change is scary, but the change also opens up possibilities, many of them now
3.
by scellus 9mo ago
My work is better than it has been for decades. Now I can finally think and experiment instead of wasting my time on coding nitty-gritty detail, impossible to abstract. Last autumn was the game changer, basically Codex and later Opus 4.5; t
4.
by scellus 9mo ago
Thermal energy storage is one gotcha. It will eventually leak away, even if the CO2 stays in the container indefinitely, and then you have no energy to extract. The 75% round-trip efficiency (for shorter time periods) quoted in other thread
5.
by scellus 9mo ago
It's complicated. Opus 4.5 is actually not that good at the 80% threshold but is above others at 50% threshold of completion. I read there's a single task around 16h that the model completed, and the broad CI comes from that. MET
6.
by scellus 9mo ago
The time (horizon) here is not that of the model completing the task, but a human completing the task.
7.
by scellus 9mo ago
I find it odd that the post above is downvoted to grey, feels like some sort of latent war of viewpoints going on, like below some other AI posts. (Although these misvotes are usually fixed when the US wakes up.) The point above is valid. I
8.
by scellus 9mo ago
I like Opus 4.5 a lot, but a general comment on benchmarks: the number of subtasks or problems in each one is finite, and many of the benchmarks are saturating, so the effective number of problems at the frontier is even smaller. If you thi
9.
by scellus 9mo ago
Opus 4.5 seems to be able to plan without asking, but I have used this pattern of "write a plan to an .md", review and maybe edit, and then execution, maybe in another thread,... I have used it with Codex and it works well. Profil
10.
by scellus 9mo ago
In other words, permanent instructions and context well presented in *.md, planning and review before execution, agentic loops with feedback, and a good model. You can do this with any agentic harness, just plain prompting and "LLM man
11.
by scellus 11mo ago
No. In general the statistics look a bit amateurish, which is normal for a scientific paper. I'd actually like reanalyze the data, just out of curiosity. (Those p-values and other things can still be on the right ballpark even if the m
12.
by scellus 11mo ago
Are citation issues related to the fact that https://www.bbc.co.uk/robots.txt denies a lot of AI, both user agents and crawlers?
13.
by scellus 11mo ago
He writes as if only datacenters and network equipment remain after the AI bubble bursts. Like there won't be any AI models anymore, nothing left after the big training runs and trillion-dollar R&D, and no inference served.
14.
by scellus 11mo ago
Not everybody, like Kevin Kelly and Tyler Cowen for example are not.
15.
by scellus 1y ago
Same here. I like the idea, have tried the social-network side a couple of times, but my kind of content is missing or I can't find it. https://bitchat.free now uses nostr for non-mesh contacts somehow, but I see no-one the
16.
by scellus 1y ago
I still use RSS, not for all content but for blogs that are now mostly Substack newsletters. Works fine, relatively noise-free.
17.
by scellus 1y ago
Even with pretraining, there's no limit or wall in raw performance, just diminishing returns in terms of the current applications, and business rationale to serve lighter models given the current infrastructure and pricing (and applica
18.
by scellus 1y ago
Yeah, just wanted to point out in my reply that the community is not paying the salaries their employees, the community in the sense of me and you, those who send observations and ids. The money mostly comes from big donors. To me that soun
19.
by scellus 1y ago
I'm not here to say it _should_ be open. Instead, I'm saying they offer a valuable, international (global) service and I want their economics to be sustainable, and have personally no objections to them keeping their AI models pri
20.
by scellus 1y ago
Not currently, but imo they _could_ sell both data products and identification to research institutions. Like having the raw data still free, but charging for derivatives, or for professional service, incl. support, related to that data. Es
21.
by scellus 1y ago
They get almost all of their money from Gordon and Betty Foundation, NSF, Nat Geo Society and the like. The "community" are almost entirely free-riders. I have donated a bit I think (maybe $20 or so), but their statistics say 0.24
22.
by scellus 1y ago
I don't know. iNaturalist contributes a valuable identification service in their free apps. I'm a relatively experienced naturalist, but greatly helped by the iOS app, simply because remembering hundreds of species names is hard,
23.
by scellus 1y ago
So far it doesn't seem like winner-take-all, and all the major players (OpenAI, Anthropic, xAI, Google, Meta?) are backed by strong partnerships and a lot of capital. It is capital-intensive this round though, so the primary producers
24.
by scellus 1y ago
The combination of Fortran and AI here starts to be close to 2001. ;)
25.
by scellus 1y ago
The new version of Bitchat (from Jack Dorsey) is interesting: it's a chat over BLE mesh, but says that it'll continue the chat on the nostr infrastructure if two (in principle anonymous) participants fave each other in the app. Ha
26.
by scellus 1y ago
People have no taste of humor. ;)
27.
by scellus 1y ago
No on AI, this is really a fringe environment of relatively uninformed commenters, compared to X. X has its craziness but you can curate your feeds by using lists. Here I can't choose who to follow. And like said, the researchers thems
28.
by scellus 1y ago
It may be a talent drain too, but at least it's a selection bias. People just get enough and go away, or don't comment. At the extreme, that leads to a downward spiral in the epistemology of the site. Look at how AI fares in Blues
29.
by scellus 1y ago
If one wants to follow AI development mostly in the sense of LLMs and associated frontier models, that's an excellent list with over half of the names familiar, to whom I have converged independently. I have a list in X for AI; it&#x27
30.
by scellus 1y ago
Maybe that would even be justified, if you are a software developer? If not now, at least soon. Imagine a software developer who refuses to use IDEs, or any kind of editor beyond sed, or version control, or some other essential tool. AI is

More ›