TechTakes

it’s very funny that OpenAI’s being so public with this, cause there’s a bunch of embarrassing shit in there:

• several admissions that OpenAI only ever used the veneer of being an open source non-profit as a recruitment, fundraising, and marketing tool — there was never a plan to function as a non-profit once they’d attracted attention
• “The core algorithms we use today have remained largely unchanged from the ~90s. Not only that, but any algorithmic advances published in a paper somewhere can be almost immediately re-implemented and incorporated. Conversely, algorithmic advances alone are inert without the scale to also make them scary.” aka, there’s no real breakthrough here, the fancy markov chain just looks scarier after you feed it the world’s data
• musk’s apparent plan was for OpenAI to implement Tesla’s self-driving tech. when they said no he threw a tantrum and decided Tesla could do this AGI thing itself. this one isn’t surprising at all but it’s very funny that musk assumed GPT was the key to making his self-driving shit work
• the general amount of cult jargon combined with clueless MBA chatter
• fucking slatestarcodex posts steering top-level decisions at OpenAI