discobot

I legitimately dont know how to reply, bc by this point llms co-own all aspects of my life and jumps between gpt4->claude3->claude3.5->o1 have all been very noticeable

the problem is that last generation of the largest models failed to overcome smaller models on the benchmarks, see lack of new claude opus or gpt-5. The problem is probably in the benchmarks, but anyway.

They explicitly train models only on public datasets

regarding the prompt processing and token generation you are correct

it makes sense to benchmark them infependently since prompt processing is done in parralel for each token and is compute bound and token generation is sequential and bound by memory banwidth

lets see how this will work out

telegram is certainly non a tech-giant with 500M users

Why is anything good objectively? It is good for me because I can stay connected with my mom despite living in different country. Also facebook group for my town is nice and useful, and my wife enjoys posting in instagram from our trips.

Now multiply this by several billions.

they enable communication for 3 billion people on the scale not previously seen in the history of mankind?

Some believe that facebook is a positive force in the world, despite being hated.

I wonder how 100K of engineers made it in

Google will happily hire and pay batshit crazy salary to any capable engineer with ML background

I wonder how this sound together with fog horns.

Should be glorious

Google search isn't groundbreaking enough?

could be related to higher taxes in Europe

Google filters the results on date of web page update, not date of the post reddit post creation, which is the correct behaviour.

And you are just making unjustified bold statements

... had very good universities with mathematics, physics, electronics, engineering. It had developed textile industry, built heavy carriers, tractors.

Well, so was Ukraine, and look at it now.

Estonia and Lithuania despite democracy and economic reforms are well known money laundromats for Russian oligarchy (see for example https://en.wikipedia.org/wiki/Danske_Bank_money_laundering_s...).

We have United Nations to handle the issues like this, and from their point of view you are wrong.

Dealing with that is CIA business, not Activision Blizzard.

Authoritarian ant antidemocratic government is just as legitimate as any other

You still need Turk to train your models...

But we have.

We do not explain how to read handwriting, we teach by example - and that is exactly how ML works.

And your last point is wrong. ML models are studied and understood much better than human reasoning.