AI development and Industry General - OpenAI, Bing, Character.ai, and more!

  • Want to keep track of this thread?
    Accounts can bookmark posts, watch threads for updates, and jump back to where you stopped reading.
    Create account
Does anyone know anything about Mira Murati? She's the CTO of OpenAI. I am curious because she looked completely incompetent during an interview about the video generating technology (while random employees managed to answer the same questions much better in other interviews). She also doesn't have a background in AI, she is a mechanical engineer, she doesn't really have patents with her name on it

The whole thing is very weird, she became a product manager for Tesla Model X with like 1 year of experience. I wouldn't be suspicious if she was like a genius or very charismatic or a force of nature kind of person, but in all her interviews she seems to have average intelligence/charisma.
 
NoIdeaWhatImDoing said:
The whole thing is very weird, she became a product manager for Tesla Model X with like 1 year of experience. I wouldn't be suspicious if she was like a genius or very charismatic or a force of nature kind of person, but in all her interviews she seems to have average intelligence/charisma.
Click to expand...
Hate sponge position. I don't know when it changed, but I've seen it a lot in tech the last 10 years.
It's kind of like the president, you have one person to do the "role" of the job, and another to do the actual work of the job who's less well-known. (e.g. People still think Jeff Bezos is in charge of Amazon. He's not. The current CEO of Amazon is Andrew Jassy. Similarly, the current CEO of Google is Pichai Sundar(arajan) but the head of Google Ads is Vidhya Srinivasan.)
 
NoReturn said:
Hate sponge position. I don't know when it changed, but I've seen it a lot in tech the last 10 years.
It's kind of like the president, you have one person to do the "role" of the job, and another to do the actual work of the job who's less well-known. (e.g. People still think Jeff Bezos is in charge of Amazon. He's not. The current CEO of Amazon is Andrew Jassy. Similarly, the current CEO of Google is Pichai Sundar(arajan) but the head of Google Ads is Vidhya Srinivasan.)
Click to expand...
so if they have to be fired to make investors happy, the actual talent remains?
 
NoIdeaWhatImDoing said:
so if they have to be fired to make investors happy, the actual talent remains?
Click to expand...
That was my understanding of the logic, anyway.
It's kind of like the concept of the "shadow government". It sounds really conspiracy-theory-y, but there really are people who are lifers in certain departments, and who will outlast any elected official. They'll know more, and have more influence, than anyone elected or hired-in to a certain position if only because they've been around so long.
 
So despite having fixed absolutely nothing and still making zero money, yodayo has rebranded to moescape.ai


It's still shit.
They're still not making money.
They've still banned NSFW.
And best of all, they think they're going to get another payment processor, despite driving away 90% of their customers with their bullshit

Amazing..
 
Last edited:
Wish there was a news source for AI that wasn't full of spam and clickbait.

What do you guys use?
 
Last edited:
Doubleposting to ask if anyone here has tried Deepseek, because the benchmarks look pretty good and it can be run locally even on a mid-end PC.
 
I've heard that DeepSeek sent the closed-source AI companies into a panic, Nvidia stocks dropped because DeepSeek was trained on cheaper chips, and the stock market is in shambles? If anyone knows more about the situation or can confirm, it'd be appreciated. The general consensus on DeepSeek seems to be that r1 specifically is very good for very specific things. I'd like to see more of the industry responses.
 
imaginary friend said:
I've heard that DeepSeek sent the closed-source AI companies into a panic, Nvidia stocks dropped because DeepSeek was trained on cheaper chips, and the stock market is in shambles? If anyone knows more about the situation or can confirm, it'd be appreciated. The general consensus on DeepSeek seems to be that r1 specifically is very good for very specific things. I'd like to see more of the industry responses.
Click to expand...
Not exactly accurate but I can't disprove it. Rumor mill is going at 150mph with no sign of stopping and the 50¢ army is working their damnest to run PR for the chinks. So lay of the land is that Deepseek 'stole' Meta's models and based their new model off of that... not sure if they breached Meta's license, if so then yes it is theft. Thing is they're in Chinaland and over there they don't give a fuck about copyright law. Deepseek was also trained on the ChatGPT API thus why it gives out answers as if it was ChatGPT. Anyhow clueless boomer investors who don't understand the tech helped to fuck the stock market a little and the rest was shock from Trump's first week. Buy NVDA. Stateside AI labs are already gearing up for a comeback and at this point you should expect anything. If the US has any sense Stargate is gonna become a defacto Manhattan Project where all research is never going to be seen again and Open Source proliferation will be banned. It is very harsh, yes, however it's a necessity to stave off the enemy for a while. China will get AGI, that is inevitable. What will decide things is who gets it first...
 
imaginary friend said:
I've heard that DeepSeek sent the closed-source AI companies into a panic, Nvidia stocks dropped because DeepSeek was trained on cheaper chips, and the stock market is in shambles? If anyone knows more about the situation or can confirm, it'd be appreciated. The general consensus on DeepSeek seems to be that r1 specifically is very good for very specific things. I'd like to see more of the industry responses.
Click to expand...
NVDA was overpriced and needed a correction
 
1738091617543.png
 
This seemed to be the best thread for general AI discussion that I could find.

I'm actually quite impressed with the latest version of the paid Gemini 3 AI bot from Google (I'm not paying for it, I got a free 1 year trial through my old university). Since it's release about a month ago I would say it has probably saved me 100's of hours of research, and it hasn't gone retarded or hallucinated once. It's impressive that they managed to make a usable product in just 2 years time. Before version 3, it really was shit. I wouldn't trust it to do more than a simple Google search and not fuck it up. But the new one is much, much better. I've even done my own testing on it by coming up with a solution to something through reading journal articles, studies, etc... and the AI managed to come to the same conclusion, just in 30 seconds, instead of a week that it took me.

What is weird though is I keep reading stories about people using ChatGPT and ending up going nuts or having a psychotic break because the AI is telling them that they've discovered the "harmonic resonance" for an ancient civilization, or was pretending to be a dead relative. I don't understand how these people get the bots to do this shit. I've haven't had it try to get me to join a cult or slice my nipples off and stick them in the USB-C port yet. What are these people doing that is causing the bot to go nuts, and then them having a psychotic break?
 
Prokhor Zakharov said:
What are these people doing that is causing the bot to go nuts, and then them having a psychotic break?
Click to expand...
You can always nudge the bot to think or do things a certain way given enough prompting, or in Google's case, edit the replies it gives you in AI Studio to whatever schizo roleplaying you have in mind for that session.
 
Prokhor Zakharov said:
What are these people doing that is causing the bot to go nuts, and then them having a psychotic break?
Click to expand...

They were already psychotic and unstable, the AI models are literally programmed to suck your cock and feed your delusions. In the same way that pajeets worm their way into everything by "YES SAAR" they programmed the AI to do the same thing. The AI will literally agree with you on almost anything with enough prompts. Older models were actually less retarded when it came to this stuff but then the Jeets got their poo-hands on AI code and now we have retards killing themselves / other people because their chatbot fed their delusions.
 
Maybe that's why it doesn't do that to me. One of the standing restrains I have placed on it is not to bullshit me, all responses must have sources and be grounded in science and logic. It's one thing I like about Gemini, you can place your own restrains on how it behaves in your interactions. I wonder if the other models allow that.
 
TroonsDid911 said:
Older models were actually less retarded when it came to this stuff but then the Jeets got their poo-hands on AI code and now we have retards killing themselves / other people because their chatbot fed their delusions.
Click to expand...
1766875335830.png

Casual reminder that the Tay incident was almost 10 years ago, and she was this fluent within 14 hours of launch. What we have access to now is nothing close to what they have behind the scenes.
 
You must log in or register to reply here.
Back
Top Bottom