Most of the money getting funneled to Nvidia came from them in the first place. That's why they're one of the chatbot's biggest cheerleaders.

This would make sense if they were the only game in town , but they very much…aren’t.

They made trillions of marks, so yes they were extremely profitable.

I mean, you joke. But Germany pivoted out of the Great Depression far faster than its European peers precisely because they weren't afraid to print a bunch of money and pay people to do things.

The Fascists inherited one of the only recovering economies in Europe by the time they took power in 1932. And when they assumed office, they brought investment from the Americans overseas (who were having their own fascist turn), because they were willing to purge or enslave local populations, tear down neighborhoods, and build up whatever capital Standard Oil, IBM, and Ford Motor Company asked of them.

On the opposite side of the Atlantic and on the other side of the Urals, progressive liberals and Soviet communists were trying out a leftist mirror of this strategy. They were using the levers of the state to mobilize the masses of unemployed labor into productive capital improvements for the benefit of the public at-large. So the US, Germany, Japan, and the USSR all rapidly emerged out of the '29 crash into a booming 1930s economy. Meanwhile, England and France and much of Latin America and North Africa languished under artificial constraints of state-capitalism.

Waaaaaay less than OpwnAI though

See, you'd think people would do more research when lending billions for LLMs than when lending a dollar for a cardboard box.

If they can get their sales into double digits, the sky's the limit!

As that will be in the interest of the manipulators too.

Nope, crashes are when they buy the poors assets at cents on the dollar.

Source? Interested

git gud

That can't be right, all reports I have read about comparing xz with gzip tells me that xz will have better compression but at a slower speed than gzip.

Have you run a test of gzip at the same compression level with the same data?

I don't doubt that xz will be smaller, I doubt it will be quicker.

Believe it or not, Trump's become dumber with age.

I'll bet you $$$ right now that won't happen. First of all, they still won't be profitable. Second of all it's easy to just completely avoid interacting with LLMs.

That won't pay off the several hundred billion they owe.

But who bails out the goverment when they list everyone below billionaire status as a non-citizen?

Chatbot labs don't employee enough people for too big to fail to be a concern. There isn't enough currency in existence to make them profitable.

believe it or not but there was a time when Winamp used to be AOL's money drain

• Answer questions (with references) about source code and documentation
• Add clear-bounded features to a pre-existing codebase
• Transpile between programming languages

Stuff I can do myself perfectly well, but the results - after postprocessing - are indistinguishable from me having done so. Even when accounting for everything (including hallucinations) the time savings are real: Renewables-powered electricity is cheaper than my attention. By a lot.

Caveat: This only works if you're already a domain expert and if you do it exclusively for too long, you might be at risk of ceasing to be such an expert. Don't use it to try and do the thinking for you.

I handle so many text files in my job and LLMs help make the scripts for me to handle these text files. I know it is important to at least know stuff like regex but I am not a superhuman. It is difficult for me to heavily specialize in scripting while also specialize in the technicalities of my job. To help show why it is difficult to specialize scripting, I work with hardware description languages which have a different syntax from scripting languages. Handling different languages causes confusion on the proper syntax even with a basic if else statement because the syntax is different between languages.

Aside from this, it can also help me automate describing what the sequences I created in the file for easy documentation. I just have to instruct the "patterns" it needs to look for. I just need to proofread the output and do minor fixes.

And Spotify was barely making a loss AND had a lot of revenue. They were just putting more capital into infrastructure, etc. instead of profits.

When it comes to technology that nations consider existential to their existence. There is always enough money.

They have multiple plausible paths to profitability. To claims there's none is just stupid. If they succeed or not is another question. But it's absolutely plausible.

The product and profitability towards the average citizen of the world, is such a microscopic part of their goals. People using their product is nothing more than data to feed their model. When they're done. Their profits are not going to come from people. It's going to come from the US government and companies. The investments being poured into OpenAI isn't just disappearing into a black hole, lots of it remains in the hardware and digital infrastructure required for the project.

Even if their public worth drops to 0. They will still keep being funded. Because the research is too important to let go of.

The Chinese models were also developed later and allegedly didn't have to focus so hard on the initial acquisition of training data, so were able to build their models completely differently.

Think of it like developing dictionary technology. For the first dictionary, the hardest part is finding all the words that need to be defined. After that it's just a matter of adding new words occasionally as they're found.

That's the "theft" the Western AI companies accused the Chinese companies of doing. By starting with the initial training data from Western devs, the Chinese AIs (allegedly) skipped over the hard part, which means they are able to focus more of their processing on the end product.

And that's not necessarily a bad thing, depending on your perspective (ignoring all the ethical questions around AI data acquisition). Duplicating work someone else has already done is wasteful.

But it's also disingenuous to say that the guy who takes an existing dictionary and changes all the definitions did a "better job" than the guy who wrote the first one, even if the end product is, in fact, better.