staunton

35 ms·

481.
by staunton 3y ago
I've never understood why this bugs people so much. I switch between languages with zero-based and one-based indexing about every hour or so. Never had a problem beyond the usual "off-by-one's" that I do anyway even with
482.
by staunton 3y ago
It sounds like the article focuses on trapping...
483.
by staunton 3y ago
I think it depends very much on the domain. LLMs are good at certain things and bad at others and these strengths and weaknesses are different from humans. The trick is to figure out how good they are at certain tasks and where it's a
484.
by staunton 3y ago
> will they ever be trustable/perfect? Humans aren't trustable or perfect either. Also, keep in mind that people strongly disagree on what "perfect" means...
485.
by staunton 3y ago
Alignment just means that the software does what you want. Of course, any term can be appropriated and abused by whoever can gain from doing so. I still think it's important to have a concrete idea what one means when one uses words. F
486.
by staunton 3y ago
My take is that the board probably never had a chance no matter what they said or did. The company already "belonged" to Altman and Microsoft. The board was just there for virtue signaling and for quite a while already had no real
487.
by staunton 3y ago
I think the ultimate jump in capability would be achieved by making a dedicated language to be used by an LLM and training it in parallel on mainly that. The problem is that to make this actually useful you need to establish a new programmi
488.
by staunton 3y ago
> you don't actually have to understand English to generate pretty decent English. You just have to have enough examples. I would have thought babies have been showing this beyond a doubt since time immemorial.
489.
by staunton 3y ago
No quality of products can compete with a monopoly. The money goes where the largest profits are expected, which is where one hopes to capture a new market
490.
by staunton 3y ago
Which also has nothing at all to do with this study by Dunning and Kruger. So you agree with the general point of parent.
491.
by staunton 3y ago
In fact, there's even more incentives for researchers to make reproducing their work as hard as possible. For example, what if someone tried to reproduce it and found contradictory results? In both cases (reproducer made mistake, origi
492.
by staunton 3y ago
By "journal space" you of course mean journal prestige that isn't unlimited. The point of science journals is gatekeeping.
493.
by staunton 3y ago
Don't even need AGI. Some idiot will do it sooner or later.
494.
by staunton 3y ago
> It's not that black and white The reason I wrote anything was that your original comment sounded very black and white to me. I think we tend to overlook how much things improve in the long term. Knowing that they do hints at poten
495.
by staunton 3y ago
It's a setup that in practice usually ends up equivalent to the nonprofit not existing in the first place. Seems like this happened here, too.
496.
by staunton 3y ago
> it will never get better Don't lose hope, other countries manage to be a lot better. It's possible if you only try.
497.
by staunton 3y ago
Would you rather be treated in a modern hospital or one from 50 years ago? Would you choose the historical hospital if it charged you the same fraction of median income as it did back then?
498.
by staunton 3y ago
Nobody will let a teddy bear manage their stock portfolio any time soon. This stuff though, I could see it happening...
499.
by staunton 3y ago
People will most definitely build such things (also into autonomous swarms of killer robots usedby the military, projects are ongoing...). However, Asimov's stories illustrate how difficult it is to find such rules. They are certainly
500.
by staunton 3y ago
Who? (if you don't want to name anyone, any hint how to find them? I personally only know one...)
501.
by staunton 3y ago
I bet you can only name one...
502.
by staunton 3y ago
What's the educated view?
503.
by staunton 3y ago
> We haven’t had unfiltered media for a very long time now. When did we have it?
504.
by staunton 3y ago
There's a fundamental difference though. The LLM is itself inscrutable, while all of these programs used to be written and understood by humans. The language used for programming used to be specified and have unique (hopefully) coheren
505.
by staunton 3y ago
It's a chaotic system, one could equally well wonder how it's possible at all, even given insane amounts of compute, especially forcasting days and weeks ahead...
506.
by staunton 3y ago
> Because of this, languages like Python are much better to use for GPT-4 coding as compared to more line-noise languages like Bash or Perl, for instance. By that argument, one should always make it use a language that's as hard as
507.
by staunton 3y ago
By that standard, almost any problem is a joke. There should be some space where something can be "no joke" without being life-threatening... It's also not an argument in the context of maintaining payment systems. Who even n
508.
by staunton 3y ago
CppCon is a very niche audience in my experience. Most "C++ programmers" use a subset of C++11 with some UB sprinkled here and there. When you care about productivity I'm starting to suspect that's actually fine .
509.
by staunton 3y ago
I don't think that's really true (though the claim might be too general to even evaluate on its merits). The same way that I don't believe most religious leaders are secretly atheists.
510.
by staunton 3y ago
How do you know? (Some anecdotes please?)

More ›