(Not the headline. The announcement. Haha)

There are many people who believe the opposite, btw. Earth has a disease. And the disease is called man. The world would be better off without humans, etc. Say that in a pub - you’ll get nods. But say you want humanity to populate the universe, and people will think there’s something wrong with you. Says less about the speaker and more about those who listens, if you ask me…

1. Pretraining 2. Instruction / chat tuning 3. RLHF

The sentence did not exist in 1 (nobody on Reddit said this, and it was also never encountered in any libgen books). It was introduced in 2 and reinforced in 3. If you stick to the base models, you’re not gonna see it (first generation only, of course).

In a sense, AI changes nothing. Society profits from having AI just as it profits from having people learning from others. In both cases, those who stand on the shoulders of others still make money for themselves. But the economy as a file is richer, too, because people can choose to buy something better now that wasn’t available before.

> In another prototype—again, heavy on the use of what he called "materials that move at the speed of thought," such as coat hangers, tape, fishing line, projectors, chopsticks, and a presentation clicker—they learned that the Minority Report style of interfacing with a computer by swiping with your hands at images was actually a poor idea. While people loved trying out the prototype, even the strongest found that their arms and shoulders soon began to hurt.

I wonder: 50ms of pink noise at the right time to make your sleep more restorative. Why or how should such a “backdoor” have evolved? Sleeping by a crackling fire? Pure coincidence, like a bug in the system that never made enough of a difference to come under selection pressure?

In effect, they’re removing attribution from the content they quote from other people’s websites. It would be interesting to see if courts object to that. It is one thing to crawl other people’s websites and display snippets of their work as your search results when each result is properly and transparently attributed. But if the text is quoted and the source is not there alongside it in plaintext, replaced only by a vague promise that, if you ask, we may or may not tell you where this piece of content is from, that is a very different deal.

Didnt Altman say a few years ago that the next frontier for training would be synthetic data? It’s what everyone turned to (Tiny Stories, etc.). Using existing LLMs to generate more training data for new ones is the obvious step.

LLMs are offered as document generators. This is what people do. They use them to generate documents. And then they train on those documents. The labs do it themselves. They didn’t ask when they trained on the internet. And now the Chinese - and researchers, and hobbyist, and businesses - dont ask when they train on LLM outputs. Especially when they paid for them.

What they call “distillation” is the very knowledge flywheel that we want in society. You buy a book, you may learn from it, and you may write a better one. The author got paid when yiu bought it. Society gets paid when the ideas spread and lead to the creation of more books, products and services, all of which increase the choices that everyone has.