pegasus

23 ms·

331.
by pegasus 1y ago
Any test?? It's failing plenty of tests not of intelligence, but of... let's call it not-entirely-dumbness. Like counting letters in words. Frontier models (like Gemini 2.5 pro) are frequently producing answers where one sente
332.
by pegasus 1y ago
We've learned a lot since then. Every time there's an accident, we learn from it and adjust the safety procedures. Incidents going down over time is to be expected, can't compare outcomes today with the past, after so much
333.
by pegasus 1y ago
That behavior is baked in the model via RLHF, it doesn't require a specific prompt.
334.
by pegasus 1y ago
Are you sure you're not subconsciously talking about something inside yourself here? Because I find your prose style equally - nay, more - annoyingly pompous than the ones you seemingly attack.
335.
by pegasus 1y ago
"Cultural assimilation" though, less so.
336.
by pegasus 1y ago
Yes, it is.
337.
by pegasus 1y ago
You could try white/pink/grey noise on the headphones, or a binaural beat generator (I use the brainwave app on iOS).
338.
by pegasus 2y ago
How are plants not lower on an evolution scale than say, mammals? They are less complex biologically and less capable of affecting their environment.
339.
by pegasus 2y ago
Why? GP's arguments seem pretty reasonable and tame. What are yours?
340.
by pegasus 2y ago
Seems to me most of that inflated budget is needed for the entertainment role of films, not the art in them, which a low budget can often stimulate rather than inhibit. In which case nothing of importance would be lost by a drastic shorten
341.
by pegasus 2y ago
I visited Knossos on three occasions and every time I returned I found it even more enchanting and evocative. Not only does it speak of a fascinating flourishing Bronze Age civilization which could be said to be, via the Greco-Roman, the
342.
by pegasus 2y ago
A better analogy would be man-hours as lumber, OS and browsers as buildings, IDEs as factories, forges or shipyards and finally end-user software as furniture and the like. As for overcoming the tension between user autonomy and developer c
343.
by pegasus 2y ago
The solution usually proposed in systems like Liquid Democracy, is in the form of an online platform where issues to be voted on are presented alongside all the different opinions and perspectives relevant to that decision. Education needs
344.
by pegasus 2y ago
Thanks for clearing that up, it really changes the story for me. This actually came up as only 4% reported by the left in Ground News' weekly Blindspot Report [1][2]. It only lists one left-leaning source, also from USA Today [3], but
345.
by pegasus 2y ago
East and West. The differences in economic output and political leanings are stark.
346.
by pegasus 2y ago
Probably some of the leftist dictators of South America would qualify. Chavez, Morales et. al.
347.
by pegasus 2y ago
I didn't think that at all. In fact I found it very readable and prefer it over the drier presentation you linked. To each its own, I guess, but there's really no need to infer ulterior motives.
348.
by pegasus 2y ago
Calculus I can understand, but the kind of coincidence GP describes is much harder to explain.
349.
by pegasus 2y ago
The program includes a lot of incidental complexity which would not be part of the spec. For example, for a sort function, the spec would be the "array sorted" postcondition. The code OTOH could be arbitrarely complex.
350.
by pegasus 2y ago
Because a proof verifier has a fixed complexity, regardless of the proofs it needs to check. Also, a minimal proof verifier is of almost trivial complexity. More advanced and performant verifiers can bootstrap from the minimal one by prov
351.
by pegasus 2y ago
"whatever shade yo want to throw at AI, 6 months from now they're going to cast it off" - like hallucinations? To me, that was and still is LLM's achilles heel. For the first couple of years we kept hearing assurances
352.
by pegasus 2y ago
Sure, you could easily say that, but it would be a pretty silly thing to say. The world is not black and white, as much as we'd like it to be sometimes.
353.
by pegasus 2y ago
The concerns you mention don't exclude the ones GP posits.
354.
by pegasus 2y ago
They still wouldn't be high quality. It's just not possible to capture the precise tone of voice in an annotation, and that precision I believe really makes a difference. My experience is that the deeper the narrator understands
355.
Software is eating the world, all right (2024)(medium.com)
184 points by pegasus 2y ago|131 comments
356.
by pegasus 2y ago
Most of the cities disappeared earlier.
357.
by pegasus 2y ago
Nobody said sharing the code "is the perfect solution". Just that sharing the code is way better and should be commonplace, if not required. Your argument that not doing so will force other teams to do re-write the code seems un
358.
by pegasus 2y ago
And not just OpenAI is facing this problem. Anthropic and Google as well.
359.
by pegasus 2y ago
But Google would do just as well (probably better) without the SEOs, no? The SEOs are just manipulating the order of search results to their own interests, but I don't see how this would benefit Google's bottom line.
360.
by pegasus 2y ago
It does work, for in-distribution detection/generalization. Which is avery useful tool for a mathematician (basically, a calculator on steroids), but it's not at all what a good mathematician ultimately does. Discovering new mat

More ›