wumbo

generating information without regard to the truth is bullshitting, not necessarily malicious intent.

for example, this is bullshit because it’s words with no real thought behind it: “if being correct was important, the developers would have taken an entirely different approach”

lmao, not what i said. fix your attention head.

metaphors grounded in reality are fine. otherwise don’t call it a “generator”

Isn’t hallucinating inherent to biological brains too?

It’s normal in small degrees even for mentally healthy individuals.

oh but think of all the “refactoring” you’re missing out on.

We have a DevRel engineer specifically working to get technical debt tickets into sprints.

So you get to refactor! Oh gee!

I think it's fair. It discerns what sounds are important, languagey sounds. It discovered but did not understand

If there’s going to be a lawsuit, go after Colt before Anthropic.

friends! let us have it both ways!

Humans have telepathy and cognitive biases!

This was already gold but went platinum with: "At the present time I do not believe there is any man of this kind."

without the z it’s entirely within the realm of English sounds.

Practice slow and I think it works.

I see it more as a modifier on the T sound at that point. Instead of letting your tongue go at the end of the T, you hold it in place because of the N.

There’s norway to prevent it

unaffjordable

“Human-level” means we can drop the bullshit qualifiers.

Human-level intelligence should adapt to many uses.

It may require teachers, but not prompt engineers.

The article author is very valid in the stance that this is far from intelligence.

Yes yes, you wouldn’t criticize a fish’s intelligence for not being able to climb a chessboard.

But these things are stupid as hell. They are not even close.

If it’s their general emotional candor, I don’t think it’s worth wondering if they are self aware.

Not your emotions, not your circus.

Why the downvotes? People are giving wrong and dangerous advice.

Yeah, you can say “well just don’t take it if it’s not prescribed.”

But don’t bury the truth. If you don’t have ED and you take it, the GP is wrong.

Severe warning. I have no sexual problems and took a viagra once.

It definitely was automatic and persistent.

It was terrible. It wouldn’t even go away when I was done with it.

I recommend a hybrid approach.

Till a new yard. It probably has chunks of concrete and other garbage from previous owners. It’s worth taking the garbage out of the ground.

Then smother the whole yard with cardboard. Then layer wood chips on that.

Dig out individual holes for plants and absolutely murder the rest of the ground with mulch.

In those holes, add compost, then plants.

With good layering, cardboard can even smother the bermuda grass menace.

if you have multiple threads working a queue of like-kinded jobs, you’ve given up the determinism of using any sort of global seed

you could pull different set lengths of pseudorandom numbers from the different seeds

this is hacker news

makes sense to me

Especially with AI. It’s better to teach the AI many examples and let it understand the implicit abstractions. Who needs to worry about reusable higher order abstractions when the machine just busts out the code you need in a single file?

Perhaps some new catalyst or meta-material would enable a more efficient horse.

If we need to grow our own food, gasoline is a problem for most American ideas of remote homesteading.

I don’t doubt the danger, but I still feel safer in a city with people to trade with than going full Walden

That’s not really an LLM at that point, but a agent built around an LLM

What would be an appropriate test?

- Given 2 word embedding sets,

- For each pair (A,B) of embeddings in one set,

- There exists an equivalence (A’,B’) in the other set,

- Such that dist(A,B) ≈ dist(A’, B’),

Something like that, to start. But would need to look at longer chains of relations.

this was genuinely my favorite video of theirs since I first saw it in 2018.

that software developer feeling

Maybe she’s attractive because she meets a very median set of attractiveness characteristics.

I.e. maybe being hot is actually less about being unique and more anout being consistent.

Maybe sultry female voices only have so much variety.

I didn’t think of Scarjo during the demo. That said, I don’t need robots to be sexy, so it doesn’t matter.

I feel the real goal is to slow OpenAI down with distractions.

Not Europe’s fault there, just determined adversarial web design.

I take it as a fair enough assumption for my judgements that all people have the same average personality potential at birth.

I’ve been around a lot of very Catholic people. I’d say half are well intentioned, whereas half are belligerent antivax etc.

The ones with good intentions prop up and obey the bad actors.

The ones with good intentions end up feeling trapped by the community and the religious trauma. Sometimes the good ones end up taking it out on their spouses/kids, perceived as units of the oppressive structure (though they are victims alike).

The difficulty of reform is all the more reason I’m happy for the slow decline in religiosity.

“some great value in religion” takes a long time of filtering through stories and metaphors to find the actual point.

I get it, some lessons are hard to teach and it’s easier to present the framework of the lesson in a story, then people make connections as they grow.

But you also have to consider all of the harmful misinterpretations that come with it.

If most people come away from religion less ethical than nonreligious people, let’s see if we can take the good parts of the texts and throw the rest away.

E.g. learning about mystical buddhism versus just going to therapy and breathing for a while.