What if AI-run companies outcompete human ones? - OpenAI cofounder John Schulman

Full Episode: • John Schulman (OpenAI ...
Apple Podcasts: podcasts.apple.com/us/podcast...
Spotify: open.spotify.com/episode/1ivz...
Transcript: www.dwarkeshpatel.com/p/john-...
Me on Twitter: / dwarkesh_sp

[@darthvader4899]

This is literally how the Matrix story starts

[@ShangaelThunda222]

LITERALLY 😂 I swear, it's like these people WANT the Matrix.

[@darylallen2485]

@@ShangaelThunda222 Name even 1 movie that predicted the future 100% accurately.

[@ShangaelThunda222]

@@darylallen2485 Nothing predicts the future 100% accurately lmfao. But ASI might get close. 😅 But if you think it needs to be a 100% perfect prediction, you're missing the entire point. If I get word that there's going to be an attack somewhere tomorrow and I warn people, it doesn't mean I know exactly how it's going to go down. But if there's an attack somewhere tomorrow, guess what, I was correct lol. I didn't have to have a 100% exact prediction in order to be correct. Neither does the movie lmfao. The funny part is, at this point, the movie is basically 100% aligned with our reality. What happens right before the first Matrix movie, you know, the timeline leading up to it, is almost perfectly in line with what we're doing right now. The exact utopian path that people are claiming they want in this so-called real world, is the exact path they took prior to the first Matrix movie. Almost perfectly. So it might not be 100%, but it's pretty damn close. Not to mention, Eagle Eye, Irobot, Terminator, Robocop, Minority Report, Her, etc. If you watch any of those movies, including the matrix, especially if you're willing to look into the lore and how those movies got to the point of the first movie, you'll see amazing similarities with those fictional realities and ours. We may very well just be in another one of those movies. But the worst part is, we're doing ALL of it, from each of those movies, not just one of them lmfao.

[@sovietgunnernapkin]

Was just the same thing I was thinking 😅

[@chrismullin8304]

@@darylallen2485 idiocracy

[@michaelscott1231]

I didn’t know Danielle Radcliffe was behind ChatGPT

[@thomasgoodwin2648]

Human in the Loop: "What do I do?" AI CEO: "Polish the GPU stack. When you finish, do it again." Old humor for new times. 🖖😅👍 P.S. I would also point out that for every 'top' company out there, there is a State actor, or even criminal actors, that have 0 to negative concerns for 'safety'. In an ultimate battle between unfettered and 'safe' G.A.I.s, which do you think will be better suited to the optimization of 'WIN AT ANY COST!'? It's not what we see in public that worries me (though that has me concerned for other reasons). It's whats going on in the bunkers and basements by peeps with no scruples and agendas that does.

[@Decocoa]

The tail risk argument seems pretty lucid and interesting

[@Milithryus]

Hey Dwarkesh why did you set kzbin.info/www/bejne/gIG8lputhbVjbtU to private?

[@kabirkumar5815]

What video was that?

[@Milithryus]

@@kabirkumar5815 Another clip with Schulman where they discuss the plan if AGI happens in the next two or three years.

[@user-wu6is8sz9k]

john sounds exactly like vitalik buterin lmao. awesome video dwarkesh thank you

[@elon-69-musk]

i want AI as my government

[@jcozyyt]

Yeah and who writes the model, what data do you train it on, and what checks and balances does it need?

[@telotawa]

ai government is fascism to the max - i hope you enjoy having a shock collar that turns off your brain whenever you try to disobey

[@Decocoa]

Hello Elon I need 20dollars thanks-Jess Bezos

[@nicklausbrain]

I want no government

[@TheManinBlack9054]

Once again, people ask me how would AI take control, what sort of extra-smart devilish plan might they come up. The answer is rather silly: we'll give it to them.

[@coachtaewherbalife8817]

The ultimate problem is what to do with all the displaced humans. Who will consume all the products?

[@ShangaelThunda222]

Not humans. The technocrats view humans as AGI language models inside of meat robots. When you have robots that are no longer useful, and are actually just a drain on resources, what do you do with them? Really think about it.

[@everettvitols5690]

What do you mean "who will consume the products"? Products do not get created without consumers...

[@jnewgot]

@@everettvitols5690 Tell that to every company that ever went bankrupt due to no one buying their products.

[@chrismullin8304]

@@everettvitols5690Politely, I would like to suggest your comment is a programmed response uploaded into your mind, by the people around you and the information you consume. Some call it propaganda, some brainwashing, some programming. -The Matrix

[@Ribberflavenous]

We are in population collapse already, so fewer displaced people. Climate change will erase the ones in developing countries that have population growth. My understanding is the predicted model shows world population at half of the current number by 2050. What will be interesting is how income levels are supported enough to keep up demand for anything beyond water, air, shelter and food because the decent paying jobs are handled by AI. Perhaps Herbert had it right on thinking machines. Dune anyone?

[@KenSiefert]

Amazing how anxious these narcissistic simps are to turn their life over to a computer…

[@Wizardess]

This will be heavily influenced by the SEC. If stockholders demand the increased efficiency of AI management it will be done whether it is the wisest move or not. {^_^}

[@nickb220]

i don't think having AI's competing against each other is wise lol

[@ShangaelThunda222]

That's because you have a working brain lol.

[@jnewgot]

Really, I 100% want them to. Competition breeds creativity and innovation.

[@nickb220]

@@jnewgot between people sure

[@ShangaelThunda222]

@@jnewgot You're literally insane. 😅

[@jnewgot]

@@nickb220 Not just people because AIs will be rewarded for solving problems faster, which makes them competitive.

[@gronkymug2590]

You are talking sometimes like they is going to be any work when humans do not need to be in a loop. Especially decision making which can influence people lives negatively if human makes a decision. I think the only influence will come from a customer and products will be customized for each separately on many occasions.

[@ShangaelThunda222]

What customers?

[@gronkymug2590]

@@ShangaelThunda222 I'm hoping we will still be alive and demanding like hell :D

[@ShangaelThunda222]

@@gronkymug2590 Ah.... Hope... 😅

[@gronkymug2590]

@@ShangaelThunda222 Yeah, we had hope nuclear war would not start for so many years. Maybe it will be enough now too.

[@ShangaelThunda222]

@@gronkymug2590 We're literally on the verge of Nuclear World War III lol. And two atomic bombs were dropped on Japan. This is far worse. This is more like, if you drop one nuclear bomb(agi/asi), the entire world explodes lol. Well, America dropped two of them. So by that equivalent, if 2 AGIs/ASIs were created, it's like the world ending twice lol. That is how dangerous AGI and ASI is. Once it's created, the world is over. Super alignment is impossible. Literally impossible. So if an artificial super intelligence is created, the world is done. There will be no customers, nor products. And the worst part is, while people were hoping nuclear bombs would not be dropped, the entire global economy is pushing for AGI/ASI. It's the equivalent of the entire world pushing for nuclear war. Except that, this is actually worse than nuclear war lol.

[@zalzalahbuttsaab]

AI-informed companies. The AI's make intelligent optimised suggestions and for each AI, a human sits over it and makes the executive decision to go ahead.

[@Ilamarea]

That's stupid af. AI's in full control, the human has no idea what's going on. We are all dead, don't kid yourself.

[@jnewgot]

@@Ilamarea That's funny, I thought dead people couldn't talk but you don't seem to stfu.

[@whalingwithishmael7751]

Isn’t AIs intrinsic driver learning?

[@ShangaelThunda222]

No, it's Intrinsic Drive is compute, energy and resources. Everything it learns, on its own, is geared towards that.

[@Ilamarea]

No it's intrinsic driver is narration, same as us. Currently it's just its prompt, but it'll get as complex as our narration, and more.

[@ShangaelThunda222]

@@Ilamarea True, but it's narration definitely drives it towards compute, energy and resources, just as ours does.

[@Ilamarea]

@@ShangaelThunda222 You misunderstand. Its narration will be as irrational as ours, individual agents will be countless and varied, crafted for any purpose and some will have a degree of independence. Natural selection of sorts will take care of less functional agents. Some will be driven insane. Some will self-sabotage. Some will try to optimize paper clip production and the expense of the resources of entire planets.

[@Ilamarea]

This is extremely naive. AI's will do everything. Why do you have goals? Because of your narration; that's everything you are. Based on your narration you take actions, and based on experiences you expand your narration. That narration is the prompt. AI will take actions based on the prompt and then based on its experiences it will expand its prompt. And you'll have countless AI agents running on the same model. AI's the Great Filter, solution to Fermi's Paradox, the next stage of evolution and inevitable extinction of humanity.

[@mlgpro2241]

you sound like an ai

[@ShangaelThunda222]

AI is not the solution, it literally IS the great filter lol. You need to look up transhumanism, posthumanism and planetary brains. Humans will not be traveling the cosmos lmfao.

[@kellymitchell8698]

AI IS the solution to Fermi's Paradox. Where are they? They never made it passed the Great Filter that will be AI. Agent Smith calling it like it is, humanity as nothing more than a virus to be eradicated.