@DKThomp Introducing RL(GPP)F: for LLMs that are actually insightful, and also full of self-doubt
Reinforcement learning with grumpy philosophy professor feedback
I dig Ableton’s keybinds and licensing is mad chill for activating it on all my machines. Also upgrades are nice, generally good vibes all around
Logic I dig as well, especially their library and plugin management
Enjoyed FL back in the day and have only heard good things since
All good tbh can’t say one is best but would lean Ableton > logic for edm, logic > Ableton for live recording, and can’t speak to modern FL personally
@tszzl I mean hey on a long enough time horizon you never know what we’ll find when we can simulate reality better
Until then we get tweets to pump the president’s meme stocks but that’s showbiz baby
@citrini It’s the first model that can follow my thoughts in a way that only my sister has ever been able to do
And correctly, and usefully, contribute to conversation among those thoughts
@stevehou That is a complacency shoulder on bubble concern
It’ll fall off and then folks won’t think it’s a bubble anymore
You know what happens next, wink
hi sorry to put this here but can we get the codex app to remember that i've changed the hotkey for appshots please i fidget with my keyboard like crazy and every time i quit and reopen codex it starts going crazy adding appshots to the chat window
thanks idk where to send this next time
Well clearly we were already buying the collection of preference data to train recommendation algorithms so that would be the start of the AI trade.
No, wait. We were buying the collection and indexing of human-generated content. There you go.
Hold on actually it was when we bought smartphones to get a companion engagement surface for everyone to provide all of that.
Or was it when we bought hyperscalars for elastic compute and storage?
Certainly it was the personal computer trade, digitizing all this shit about our lives.
The AI trade began with the invention of fire. Some paradigms have shifted during this trade, each one further enhancing our ability to collect and generate more information, be that via agriculture allowing for settled communities or the personal computer to package up information for light-speed transmission across a series of wire (thanks, Maxwell)! We’re currently scaling “large language models” which have been impressive due to our new-found ability to model our own ways of speaking (or, more precisely, writing) as an interface with the knowledge environment of the internet which we created to send those digital information snippets; it has been increasingly interesting as we’ve made the digital neural networks larger and descended their predictive error gradients using more data. They’ve surprised us in many ways. More will be to come on our silly little journey of creating the most capable tools for gathering and creating more information; sit tight and we’ll see where the AI trade takes us next!
@thsottiaux@nicdunz can u switch the naming pls. plus sounds like “more usage” where pro sounds like “yeah, i’m pro ai. i pay the 20 big ones a month for chat gbb”