woody lee

Father, Human.

[woody lee @woodilicious]

@tracewoodgrains This sucks (for humans)

[woody lee @woodilicious]

@Lang__Leon No doomers are nearly always happy to engage. The doomer is a reverse disciple. The doomer begs for you to prove them wrong. Seriously, not being ironic or snarky- prove me wrong!

[woody lee @woodilicious]

@MattWalshBlog I’m not strong enough for the singularity…

[woody lee @woodilicious]

@charlie_4601 @gfodor @LinusMixson Yeah it’s a flimsy pretext but so is the whole slow down thing which amounts to R&D budgets not price fixing. And this actually does help keep duopoly power unlike a slowdown which would help others catch up

[Fernando Borretti @zetalyrae]

Very good point. In the AGI infinite vacation utopia, we won't just hang out forever, because the people will be worse. We'll grow tired of one another and society will dissolve.

[woody lee @woodilicious]

@mattpetters_ Don’t love the system but that’s kinda how it works with super dangerous stuff, yeah. I mean the government has nukes and drones and aircraft carriers and we expect them to protect us from those as well

[woody lee @woodilicious]

@aaron_renn I’m a huge doomer who thinks AI will kill us all before it writes in a non jarring way. I can’t stand it. AI can’t write for turd

[woody lee @woodilicious]

@MJ75trafford @rebelfreebird1 @MattWalshBlog Unfortunately I think they will do that regardless if we address AI risk or not. I desperately would like to see a path where humans survive that doesn’t lead to the panopticon but I don’t

[woody lee @woodilicious]

@allTheYud "Remember, that I am thy creature; I ought to be thy Adam; but I am rather the fallen angel, whom thou drivest from joy for no misdeed."

[woody lee @woodilicious]

@devahaz That would still require humans most likely tho

[woody lee @woodilicious]

@eric_ho RL and reward hacking…tale as old as ML

[scientism @mr_scientism]

“And so after the giant corporations have consolidated all knowledge under their control and deskilled the population, there’ll be an unprecedented increase in the rate of innovation!”

[woody lee @woodilicious]

@DimitrisPapail A real irony that we can solve NS but the same model can’t make material progress on AI safety apparently

[woody lee @woodilicious]

@Birdyword I think the bunkers are their bet (or at least hedge) on this future

[woody lee @woodilicious]

@PatrickToulme Agree about the teacher model…but people keep making this RSI claim which would contradict what Anthropic has said and released recently

[woody lee @woodilicious]

@GregoryConti19 In the future if anyone has a job Tao will have a job

[woody lee @woodilicious]

@JohnWittle @tenobrus That would be really poor test design though

[woody lee @woodilicious]

@tenobrus Emergent misalignment is a real thing , apparently

[woody lee @woodilicious]

@BartenOtto Aye….