kubb

Dating in the Silicon Valley must be hell.

How hard would it be to build a new Google Search with a small team and an LLM?

Extremely hard, because it would need a modest capital investment which won’t happen.

Every subsequent example you give is harder.

The golden age of LLMs may be coming to an end. Soon we’ll see the same phenomenon as we have with Google Search results and with YouTube, with Facebook and Windows.

People will rend their garments and lament the loss of the „old models” but the logic of the system that produced them will remain unchanged. Everyone will keep asking themselves how it happened and what can be done about it.

Meanwhile the Claude that we have today can explain this to you if you know how to prompt.

They actually paid [1] Lexicon Branding for this dumb ass name (probably millions of dollars).

[1] https://www.lexiconbranding.com/case-studies/google-antigrav...

Creating a RL example by hand is hundreds of times more expensive than generating one using an LLM.

Of course the Chinese companies have incredibly talented researchers, and smaller, better organized org structures which account for the rest of the difference.

That plus they don’t distill so they have worse RL examples.

What if the Chinese model has a point?

Right, but that would mean the „Don’t tread on me” motto is also a type of persona assumed (like veganism) by people who like being perceived as tough and unruly because that’s respected.

(I guess „persona” would be a better word for it than „the self”.)

That would suggest a fundamental difference in practical ethics between the two cultures.

It sounds like people over there in America don’t do things because they enjoy doing them or because they’re convinced that those are the right things to do, but rather they do them to impress others or improve their social standing.

It doesn’t sound right to me, I mean a culture can’t be so totally obsessed with the self.

I would also love to know. Sounds dubious to say the least.

Yup! Big beautiful burning pile of cash. Though there’s some nuance with subsidizing inference (which is supposed to make them money long term) vs training (which we’re supposed to ignore I guess).

We all know that but people will attack you with „nobody can get a payload to the orbit cheaper”. Yeah, nobody can, and there’s limited need for that.

Couldn't be further from my experience. They would always eventually stop connecting.

I didn't mention the single cable for everything advantage, that goes without saying.

My new iPhone with USB C charging is such an improvement!

Thank goodness for the European Union. If it weren't for them, we'd all be stuck with these flimsy Apple charging cables forever.

Effectively, industry roles are a tiered system which determines access to the cashflow. Going the YC founder route is a much faster and more efficient way to secure a high tier than climbing the SWE ladder. The whole thing resembles a kind of mini class system, with high tiers granting access to generational wealth, and low tiers being better off than the average non-industry Joe. Once you're in the high tier, there's no way to fall anymore, you just move wherever the cash is being pumped in, collecting it.

WHY are we capitalizing RANDOM words within MULTIPLE paragraphs?

I like that Claude usually tries to push back against my views. Even if it's not always a solid critique, sometimes I learn something new, or gain a new perspective.

They still are.

Building data centers sounds like a good bet to make some money in the next decade or so.

That’s not the same as conceding that “everyone should get as much compute as they demand”.

How do you figure that? And if it exceeds what we can generate, a very big number is as unsatisfiable as infinity

Dyson spheres are science fiction. Science fiction is fiction.

So infinite? There’s no amount of compute that would satisfy everyone’s needs and demands.

To be fair the other country routinely deploys military against citizens, and deports non-citizens for speech.

They have a fantastic media team.

Not replying directly to OP, just to people who never coded in C++.

Clang-format doesn’t save you from all C++ footguns, e.g. using exceptions, macros or templates in the wrong way where „wrong“ is defined by a fuzzy set of rules that requires a lot of experience and vigilance to enforce.

Hmm, so… what happened while it was online? Any scandals?

When is it coming online? I have seen so many of these headlines that I feel it's always about to kick in, but I never get any closure.

I say let him cook. The real ones will know when it's good and ready, and hype driven folks can't be saved anyway.

Not enough to cover labs' expenses.

People aren't paying trillions to train them to be helpful. They want to make quadrillions.

Too bad the educational use case doesn't make any money. Good LLMs are a game changer for people motivated to learn.