2025 models will be more like coworkers than search engines - OpenAI cofounder John Schulman
Ғылым және технология
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
Пікірлер: 48
Just drop the episode already 😭
An AI that I would be able to share screen with and talk to in real time through voice as if in a zoom call would be the best device I could ask for.
@forthehomies7043
14 күн бұрын
OpenAI just did a demo presentation for their newest 4o model. It’s on KZread, you’d love it. It’s exactly what you’re looking for, real-time voice comms and live video sharing and more.
@user-jd3us8io5e
14 күн бұрын
It will be your assistant until it replaces you. The entire purpose of AI is labor replacement. It will start with enhancement then turn to replacement. I am amazed at how few people seem to see this as an issue. Given that the distribution of wealth has never been even, I expect this to destroy the middle class. The companies using AI will profit and there will be a lot of permanent unemployment.
@CYI3ERPUNK
14 күн бұрын
XD well ur in luck cuz thats here now as of yesterday
@cagnazzo82
13 күн бұрын
My dream is having an AI that you can watch a movie with and have it making funny comments or pointing out historical facts. OpenAI's 'Her' is already partially heading in that direction. She was pretty funny.
@jackfarris3670
13 күн бұрын
@@user-jd3us8io5e I have credentials needed to sign off important financial papers. It will be a very long time before the government allows AI to replace a job like mine (health care industry/ heavily regulated). Even though it could do my job some future day, it will be even farther when a company can logistically replace me.
"Coworker", or the real word. Replacement. What is a coworker, other than "just another worker"? Sure, it'll be incremental, but it's obviously going to be incremental by the month, not the year.
@cagnazzo82
13 күн бұрын
The debate is whether you will be replaced or whether you will be augmented.
@ev.c6
9 күн бұрын
The coworker in this case is not proactive. You need to tell it what to do every single time. This is not AIG.
@petergraphix6740
8 күн бұрын
@@ev.c6 You should see how many of my coworkers are not proactive....
There is no one else in the scene that knows and asks the right questions like you do. Even better, you actually go to the best possible sources to get the best possible answer. Thank you.
@ironeagle4274
13 күн бұрын
I haven't seen much of his content yet, but I'm new subscriber for one reason. He asked pertinent questions that I felt showed keen awareness of what he was talking about. I'll definitely watch more now that I have seen him preforming this well for a second time.
Why drop the short-form content before the long-form?
@leastofyourconcerns4615
14 күн бұрын
ask those who came up with trailers
@_ProveIt
13 күн бұрын
Don't hate the player, hate the algo
@Greg-xi8yx
13 күн бұрын
Yeah it makes no sense. 😅
Dwarkesh, always Great topics
Great topic!
This is like advanced politics to talk in such reassuring terms when the easy reach for truth is far more surprising.
When I spend time with gpt-4o I still find that it has biases and or part of conventional wisdom, i.e. not true. My fear is that bad ideas will be locked in for decades.
@ssehe2007
12 күн бұрын
For example?
Hey Dwarkesh great podcast! Can you please please get David Kirtley from Helion on? Given the hype around fusion and Altman's backing of him, it would be a treat!
This will end horribly.
Humans are not that good at long-horizon tasks. But this might be a good thing. Iterate, drop things that aren't working, try different stuff. Interrupting long horizon execution / validation will become an even more important problem. As timelines grow, costs of mistakes will grow too.
Just like AI, John Schulman got released by OpenAI and now we have a cofounder, interesting
I like keeping pace with the advancements in LLMs, but I don't want to use them, creeps me out. I kinda hope global society implodes and we stop working on this whole thing entirely. Just seems like we don't need it and the dangers are too great to risk it. I know that's not likely, too much momentum, just seems like we're accelerating towards something we don't understand for benefits we don't need.
@MrNote-lz7lh
13 күн бұрын
Too bad. So sad. The industrial revolution ended slavery. The robotic revolution will free people from work, make them immortal and give them space travel and FDVR. And there's nothing you can do to stop it.
Nukka named Dwarkesh. 🤣
Lmao he's looking to his right every few minute or so, probably someone from the company is standing there with a gun in case he spills something :).
lfg
Yeah, I'll believe it when I see it. Copilot has so incredibly over promised and under delivered. That seems to be the motus operandi for all the AI companies.
So make search engines suck, (they are a shadiw of what they were 5 years ago) then make ai as good as a search engine from 5 years ago, then say it a cowerker? Great plan.
@TheBoogerJames
12 күн бұрын
Seach engines do work. It's just that the work they do is making money for the company, not showing you relevant search results. 😮
This guy reeks of social awkwardness. No wonder he's the cofounder of OpenAI.
@BritishProudnShit
14 күн бұрын
Hmm, don’t really see that. What exactly is off for you?
@jonathanlucas3604
14 күн бұрын
He's accomplished more in his single life than you, your ancestors, and all of your descendants. No wonder you resort to insults.
@AmericazGotTalentYT
14 күн бұрын
So do you
@jonathanlucas3604
14 күн бұрын
@@AmericazGotTalentYT oh I reek of awkwardness too, I don't deny it. I also have accomplished more than you or your descendants. Goodbye
@telotawa
14 күн бұрын
normies trade intelligence for social skills autism remains peak intelligence