Viewing profile — scellus

the main bottleneck for the civilization in communications currently is the sparsity of cynical, negative HN comments

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 t…

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…

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 …

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…

The time (horizon) here is not that of the model completing the task, but a human completing the task.

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 usua…

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 nu…

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…

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…

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 …

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?

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 tr…

Not everybody, like Kevin Kelly and Tyler Cowen for example are not.

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-…

I still use RSS, not for all content but for blogs that are now mostly Substack newsletters. Works fine, relatively noise-free.

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 gi…

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 …

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…

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 p…

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 thin…

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 becau…

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…

The combination of Fortran and AI here starts to be close to 2001. ;)

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 anon…