They should have known better than to train it on all that new age fakery from the 60's, LOL.

(But they didn't. Because we're dealing with very stupid smart people.)

Not letting you into my house anytime soon.

I call it taro. But Purptato works.

For anyone wondering on how I currently conceive of this:

LLMs - senses (percepts), knowledge storage (concepts), and memory access (summary, application)
Reinforcement Learning - reasoning, prediction
Synthesis - new concept formation

Feel free to point out if I'm way off base.

Sam Altman's midwittery in over hyping LLM's is going to severely damage true AI research for quite a while when this bubble pops.

He's portraying 1/3rd of a trip as the entire journey. And that will make us less likely to take the next steps. We won't want to be suckered again.

For synthesis and innovation, that's a lot harder. What enables the flash of insight we gain from connecting two ideas of relatively low association? Modeling this on a computer is roughly equivalent to infinite monkeys with their typewriters. (Which is how LLM's are now.)

For learning, reinforcement learning seems best. We saw research on this in the early days of AI with trying to have computers replicate animal learning algorithms in real time. Human learning is more complex.

While we are not blank slates, a computer would be. But the LLM component enters there.

This, however, suggests you could get to AGI if you had engines that perform the other two tasks and relate to LLM's. LLM's replicate our ability to spot patterns based on the concept data we have.

It's necessary as a component to replicate our intelligence. But it's not sufficient.

Palantir trains its LLM on the vast stores of intelligence data gathered by the NSA and other agencies. They make significant money from government contracts. Since they have an actual revenue stream, they will be able to weather the bubble pop.

Also made some small tweaks to the schedule template. Testing some curvature tricks for effects.

Superb movie. Caught it the other night. Lived up to all the hype.

Main villain for System Shock.