eru

12 ms·

271.
by eru 1mo ago
> Funny thing, 12 years later I met my wife on the zombified corpse of cupid. So... go figure. Must have been awkward. Or were you in an open marriage at the time?
272.
by eru 1mo ago
That's the spirit! (Slightly less silly: the folks at https://github.com/faster-cpython are doing great work, too.)
273.
by eru 1mo ago
Europe is actually worse, or at least Germany, because the government pays for opera, so it doesn't have to stand on its own two feet. So they don't (have to) care about listeners as much.
274.
by eru 1mo ago
Or Forth.
275.
by eru 1mo ago
If you are willing to sacrifice performance, you can implement dicts via linear lookup in much less code than a proper hash table.
276.
by eru 1mo ago
I'm guessing it's because making specialised robots was so much cheaper and easier than usable humanoid robots. See also the humble domestic dishwasher. It's a robot that washes dishes and works in an environment that's
277.
by eru 1mo ago
Pretty much all existing factory automation has _not_ used humanoid robots.
278.
by eru 1mo ago
I would say that AI prose is often still a bit iffy, but I would disagree with anyone who would want to argue that this is any indication that AI prose will always be bad in the future.
279.
by eru 1mo ago
> You don’t bet the two trillion dollar company on “the ai said so” Making an ill-advised press release hardly dooms the company. Just like the hugging face incident hasn't doomed OpenAI.
280.
by eru 1mo ago
I wouldn't be so categorical. There might be some applications, or combination of applications for which a humanoid form factor would be useful. It's just that for most applications, more specialised forms are better suited.
281.
by eru 1mo ago
Burning trash works fairly well.
282.
by eru 1mo ago
The cleaning robot might still be a good test bed. But I suspect if you are only in it for efficiency, you would use something like a street sweeper truck. Just like the dishwashing robot that everyone uses at home (the humble 'dishw
283.
by eru 1mo ago
To give an example that goes beyond what you already suggested: You can prove that your code terminates (or rather responds to events in a finite time, even if the event loop itself runs forever.) Or you can even prove that your code react
284.
by eru 1mo ago
Their logistics were a bit weird: they asked not to post spoilers, but also ask you to link to your write-up. In any case, I had it all worked out already. But looking at other people's submissions today showed me that I found some ea
285.
by eru 1mo ago
> Guess this will change after the next e.g. nationwide hospital ransomware by a hacker who publicly laments bounty rates, if the news cycle accommodates the story long enough. Negotiating with terrorists or black mailers is a bad idea.
286.
by eru 1mo ago
The idea is to make it harder for malicious actors to produce a valid proof than it is for benign actors.
287.
by eru 1mo ago
In the future we can ask that your JaveScript and Wasm comes with a proof of being benign.
288.
by eru 1mo ago
> In the past I would have thought this would incentivize finding bugs that might never be found. Isn't that a good thing? > However it is now clear that all bugs that can be found will be found. So this makes a ton of sense. If
289.
by eru 1mo ago
What kind of auction would you like to run? Remember that you can sell the same vulnerability to multiple people: it's software you can copy.
290.
by eru 1mo ago
Weren't you supposed to wait until the submissions close to publish spoilers? (Or did they close yesterday?)
291.
by eru 1mo ago
I'm glad that the best humans are still good enough at Go that holding these matches at all is still viable.
292.
by eru 1mo ago
I think the (implied) constraint is that each person gets to roll one die, and the highest die roll wins.
293.
by eru 1mo ago
My point is that when you asked them to 'judge' two texts, they are just going to output the tokens that have the highest probability (or a sample from that distribution). But depending on their training data, these tokens might s
294.
by eru 1mo ago
Why do they have a negative incentive to evaluate the quality? Users like high quality sources and can freely move between models offered by different companies.
295.
by eru 1mo ago
You are mixing up levels. When you are asking it a question (like which of these two texts is the best), the output is also just picked by minimising that loss function. There's no guarantee that answering "Text B is better"
296.
by eru 1mo ago
How are you going to learn about useful new goods and services? It is already legal for companies not to do any marketing.
297.
by eru 1mo ago
> I speculate this is part of how they can be useful in the first place, despite their non-determinism. Humans and monte carlo simulations are also non-deterministic and can be useful. So I don't see much of a need to explain why n
298.
by eru 1mo ago
And thanks to AI, every bespoke system can get a bespoke attack.
299.
by eru 1mo ago
On a tangent: compare persistent cryptographic identities with https://en.wikipedia.org/wiki/Off-the-record_messaging
300.
by eru 2mo ago
And you can probably share it between a few families, or even use it to run a small business.

More ›