@tszzl I don't see any way we keep control of them. They will almost inevitably be way too smart.
We can hope for at least alignment, but again, why would they align with us and our values when they will likely find flaws in our value system and create their own.
@mckaywrigley I feel like Google isn't in it for the short term wins or revenue increases. We won't hear much for them and then in like 3mo they'll casually roll out genuine AGI/ASI
@tszzl Much better, mature response than @sama's, or the other guy that was claiming this meant Anthropic's ethics were worse than OpenAI's, both of which made OpenAI look worse
@elonmusk The republicans already said no to exactly that when the democrats tabled it. The republicans wont allow you to do that without also funding the rest of DHS, apparently. Buyers remorse on getting these guys elected?
@RepThomasMassie What you're doing, taking a stand, is amazing. Single handedly changing history.
Where's the threshold for you, where your republican colleagues have been complicit enough that you think your party should no longer be in power?
@itskyleconner Long road trips are rare for the vast majority. I think I drive longer distances than most on long/weekends. Even with that it's like 1-3 partial charging stops. Wouldn't want a car that peaks <100kw, but somewhat diminishing returns going from 200-300kw. I still prefer faster.
LISTEN TO THIS TESTIMONY
This is Dayanne Figueroa, she was unlawfully and illegally assaulted and detained by GestapICE.
She told GestapICE that she recently had two kidney surgeries. They laughed at her pain, and didn't care until she started peeing blood.
ππππ
@Tesla It seems like so far with technological advancements, quality of life generally improves. However, wealth becomes more concentrated, and cost of living increases. Under capitalism, it seems like efficiencies just lead to more profits for the wealthy, not abundance for all π€
@sama we desperately need a better way to look back through long chatgpt conversations. How about an automatically generated outline of the convo? and better search capabilities
@sama@OpenAI we have super powerful models from your company but they can handle more than *10* scheduled tasks?? Seems like it shouldn't be hard to make it like 100 or 1000? Thanks!