@zetalyrae Agree that property rights including AI company equity may not be enforced in the future, but disagree that humanity will be useless.
Humans (at least the ones who try) can be surprise machines that sample out of AI's distribution.
@iruletheworldmo You forgot purgatory. Long term will be a mix of heaven (high-agency, empowered people loving every day of building) and purgatory (low-agency or disempowered people living on dopamine drips).
We just crossed 10k GitHub stars!⭐
None of this happens without the open source community and the K-Dense team. Thank you. If you have starred, contributed, or are building with us, this is yours. https://t.co/mYIM0qs5sf
@DaveShapi Hi @DaveShapi , curious what you think about my hypothesis that China actually *is* building a Manhattan project for AGI: https://t.co/zEnnoLSqIE
@TheGeorgePu What was the acquirer buying? In acqui-hire cases, they value the founder, and will give a nice multi-million package to retain him for a few years. In investor soft landings, acquirer is doing the investor a favor and the founder got minimal face-saving money.
@biostateai P.S. Even when we requested p=49.9999% chance of returning "1", it still almost always gives us "0", so it's not a simple logits quantitative issue.
What does a scientist do, and can AI do it?
Broadly, science can be separated into 2 steps:
Coming up with an idea to test, and testing/validating the idea. Step 2, the validation, requires the scientist to be careful and accurate, which current LLMs are good at.
But Step 1, idea generation and selection, requires exploration off the beaten path -- and this is something that all LLMs currently struggle at.
In Biostate’s latest research, we show that LLMs show step function-like behavior for a binary decision: meaning if we ask for ChatGPT to respond with “0” 49% of the time, it actually responds “1” over 95% of the time.
We still have quite a bit of work to do before AI can do revolutionary rather than incremental research!
https://t.co/Pl8Bxa5JKS
@kunalb11 This is almost certainly tautologically true if what you mean by interesting is "watching a train-wreck of a life". On the other hand, I think Warren Buffett is very interesting.
@PeterDiamandis Why not the equivalent of High Intensity Interval Training? E.g. every 5 years, work 5 months at 16 hours a day for 7 days, and take the rest off for vacation. Work hard play hard.
@JuliaEMcCoy Support your optimism! Realistically, I think it will be abundance-agency balance. Every person at every moment will have the option to live like kings/queens on their preferred stream of dopamine, or choose to struggle and do hard things.