ACCount39

19 ms·

1.
by ACCount39 6d ago
That's the thing: we ripped the entire idea of learning connectionist networks straight out of the brain - but our implementation of that idea is very different. The asynchronous nature of BNNs? The brain does that because its neurons
2.
by ACCount39 6d ago
The AI industry in Russia is not doing well. Domestic AI models exist, but they aren't exactly high performers, and many of them are based on the open weight Chinese models - so there is very little reason to use them. IT power users r
3.
by ACCount39 7d ago
Multiple jurisdictions have tried to play the stupid games of "we'll make a dumb fuck law that creates problems for you, the website operator, and if you don't conform, we'll sinkhole you at ISP level". Platforms li
4.
by ACCount39 7d ago
Agreed on both counts. 20 MHz is well within the range of any random scope, and you should have at least one "any random scope" on your desk if you're doing electronics seriously enough to reach the "diagnosing QA fail b
5.
by ACCount39 7d ago
"The brain does it" is not in itself a reason to do the same in AI. Sometimes, what the brain does is genuinely a good solution to a given task - one that's good regardless of whether your neural network is of artificial or b
6.
by ACCount39 7d ago
Thanks for the link. This is the kind of "edge AI" adaptation that I expect to see more of in the future. A "local 8B VLA" that runs at high frequency can be sufficient for things like sensor fusion, plan execution and m
7.
by ACCount39 7d ago
Yes. That's the issue: we are looking at the same posters. It's just that "seeing this poster for the first time, unaware of the workings behind it" is a very different experience from "seeing this poster after you
8.
by ACCount39 7d ago
It's saturation. If I saw an Images 2.0 flyer in year 2020? In a world where Images 2.0 didn't exist? I'd be impressed. But if I see it now, today? I'm not impressed at all. It's the same image, produced by the same
9.
by ACCount39 8d ago
It's an old paper (from 2025, so, a decade ago in AI years), but the concept is still fascinating. And I'm yet to see it show up in any production models. If multiple models can use cache representations for this kind of enrichmen
10.
by ACCount39 8d ago
"Just need some breakthrough materials to work" is the kind of "just" that stops things dead in their tracks. So far, the game in town has been primarily hydraulics or electromagnetics, and the things that replace them a
11.
by ACCount39 11d ago
Oh, if it isn't the MSM8916 himself! Fancy seeing you out there in the wild - posting on HN no less! Is there a guide for cheap MSM8916 platforms you would recommend? There is a hkfuertes repo, but the one I found is archived. Not to s
12.
by ACCount39 11d ago
What it has at its core is a smartphone chipset from 2014 - MSM8916 or so. Not exactly an AI compute powerhouse.
13.
by ACCount39 11d ago
That was genuinely a strong signal that whatever learning algorithm the brain uses isn't "magic", and probably can be approximated with the ML tools we have. It also pointed at the possibility that the learning algorithms bra
14.
by ACCount39 12d ago
You picked a piss poor metric and you are getting piss poor results. You can absolutely train an LLM-based model to be superhuman at chess. We just don't care enough to do so. LLMs being even as good at implicitly tracking chess board
15.
by ACCount39 12d ago
If chess ELO is the best proxy for general intelligence, then surely, even Deep Blue was smarter than any human alive? You can also claim that the best proxy for general intelligence is being able to multiply large numbers. In which case bo
16.
by ACCount39 12d ago
This. They've been saying "AI is an extremely powerful, extremely dangerous technology" back when actual AIs were image classifiers that would struggle to tell a cupcake apart from a dog. OpenAI was founded by people who didn
17.
by ACCount39 12d ago
It's very easy to tune an LLM for a "default" like "be sycophantic" or "be contrarian". It's easy to instill a semi-rigid "response template" like "agree with most of whatever the user
18.
by ACCount39 12d ago
The successes are in areas where you can get AI to execute an iterative loop autonomously. Try, fail, refine, retry - succeed eventually. Coding is, inherently, quite amenable to that. Is bioscience? They're certainly working on making
19.
by ACCount39 12d ago
Xitter can't make a web UI that's actually usable to save their lives - so they're going after the one platform that does with lawfare. Despicable behavior.
20.
by ACCount39 12d ago
The issue is, you kind of have to not get wrecked by the negative impacts to collect those positive impacts. Unfortunately, "AI that's competent at bioweapon R&D" might be an easier mark to hit than "AI that can put
21.
by ACCount39 12d ago
Frankly, I would be more concerned if they didn't. Given what we know of how unaligned are they in practice? All the proto-Astra and Mythos incidents? Not hacking there would point towards pure benchmaxxing/eval gaming.