News Ex-Intel CEO Pat Gelsinger's startup chose China's DeepSeek instead of OpenAI

[Admin]

Pat Gelsinger credited DeepSeek's creative engineering for driving competition and making AI more affordable.

Ex-Intel CEO Pat Gelsinger's startup chose China's DeepSeek instead of OpenAI : Read more

 

[hotaru251]

Gelsinger now serves as chairman of his new startup, Gloo, which is reportedly a messaging platform for Churches

...why does a church msg platform need ai????


also

DeepSeek is under investigation by Microsoft and OpenAI for allegedly using data from ChatGPT illicitly in distillation.

because every "ai" did exact same thing? you an't call foul play when every one of them didn't pay for the content used to train the systems at start.

A thief cant report being robbed of what they robbed to begin with.

 

[SonoraTechnical]

man... 63 years old and worth an estimated 85 million dollars even after the intel stock price slide. He has 8 grandchildren.... retire man....

 

[PBme]

man... 63 years old and worth an estimated 85 million dollars even after the intel stock price slide. He has 8 grandchildren.... retire man....

Seriously. Don't get rich older folks that are like "yeah, I want to slave over a company 24x7". May be the same obsessive traits that got them to exec positions but logically seems dumb. Dude could drop from a heart attack any time. Even if he makes it, you have maybe 10-15 years left of enough mobility to do a lot of types of travelling. The world is rapidly on its way to burning (figuratively and literally). 1000 reasons to maximize the little time he has to enjoy being on this earth and this is what he wants to do?

Also, seems laughably stupid to base anything in your company on a chinese AI product. Countdown to it being banned in 3..2...

 

[King_V]

DeepSeek is under investigation by Microsoft and OpenAI for allegedly using data from ChatGPT illicitly in distillation.

because every "ai" did exact same thing? you an't call foul play when every one of them didn't pay for the content used to train the systems at start.

A thief cant report being robbed of what they robbed to begin with.

My thoughts exactly? Isn't this the very same OpenAI whose CEO said they NEED to be be exempt from any laws regarding copyright or intellectual property, or they won't be able to make it?

That's a lot of chutzpah for OpenAI to be part of this investigation.

 

[soclibfiscon]

Must have been a pretty short investigation by Microsoft. I just saw an article saying that MS is hosting deepseek on Azure.

 

[phead128]

OpenAI has stolen data from newspapers, publishers, artists, everywhere, so now crying that it's data is trained on, hilarious.

 

[evdjj3j]

man... 63 years old and worth an estimated 85 million dollars even after the intel stock price slide. He has 8 grandchildren.... retire man....

Work is a lot different when you're one of the ones at the top.

 

[Deepwinter]

Seriously. Don't get rich older folks that are like "yeah, I want to slave over a company 24x7". May be the same obsessive traits that got them to exec positions but logically seems dumb. Dude could drop from a heart attack any time. Even if he makes it, you have maybe 10-15 years left of enough mobility to do a lot of types of travelling. The world is rapidly on its way to burning (figuratively and literally). 1000 reasons to maximize the little time he has to enjoy being on this earth and this is what he wants to do?

Also, seems laughably stupid to base anything in your company on a chinese AI product. Countdown to it being banned in 3..2...

He clearly does it because he enjoys Working, it probably keeps him in a healthy routine, my dads in his mid 70s he still goes to work every day, leaves the house at 7am and leaves the office at lunch time.

I'm sure he's not using deepseeks API he's most likely running the model on his own hardware, or using an American company's version of deepseek R1
It's already available on grok. Not the twitter Groq but the actually Ai company called groq.com
They are 20 X cheaper per input token and 60x cheaper per output token.
Based on openais per million token API calls same for Groq.com

In fairness Groq is using a distilled version of 70 billion parameters but it is showing similar results to the 671p version.

I suspect open ai will be forced to implement many of the innovations demonstrated by deepseek into there next AI, because the price difference is to much now.

 

[christopher.andrew.carr]

" ... superior performance compared to OpenAI's best .. "

No, OpenAI 's best is o3. It is not superior to that. It's about on par with o1. And it's neck and neck with Gemini 2.0 Flash Thinking (but Gemini is much faster) -- and Google has the larger base model 2.0 Thinking coming soon.

People need to chill, and ask themselves if they actually know what they're talking about in this topic before they hit "enter."

 

[King_V]

People need to chill, and ask themselves if they actually know what they're talking about in this topic before they hit "enter."

Excuse me, but this is The Internet™, and having to know what you're talking about before posting something? Well, that's just crazy talk! 😅

 

[Constellar]

Guys--some of you are complaining because OpenAI is (justifiably) taking legal action against DeepSeek because, and here's your reasoning, the latter effectively stole training data from the former because OpenAI "also stole copyrighted data for their models to train on".
I'm surprised people don't see the obvious flaw in this line of thinking. "Two wrongs don't make a right," or however you want to phrase it, by stealing from OpenAI, DeepSeek has effectively committed theft not just once, but twice: first, by stealing from OpenAI DeepSeek is automatically stealing any copyrighted data that OpenAI acquired, and secondly through DeepSeek's direct theft of the OpenAI corpus itself.
I'm not sure where the source of everyone's cognitive impairment lies; that is, I'm not sure why it's proven difficult for people to understand that two wrongs don't make a right. All I can say is that we must support OpenAI with all our energies, since they have the proven track record of innovation here, not some substandard and cheap product like DeepSeek...

 

[phead128]

Guys--some of you are complaining because OpenAI is (justifiably) taking legal action against DeepSeek because, and here's your reasoning, the latter effectively stole training data from the former because OpenAI "also stole copyrighted data for their models to train on".
I'm surprised people don't see the obvious flaw in this line of thinking. "Two wrongs don't make a right," or however you want to phrase it, by stealing from OpenAI, DeepSeek has effectively committed theft not just once, but twice: first, by stealing from OpenAI DeepSeek is automatically stealing any copyrighted data that OpenAI acquired, and secondly through DeepSeek's direct theft of the OpenAI corpus itself.
I'm not sure where the source of everyone's cognitive impairment lies; that is, I'm not sure why it's proven difficult for people to understand that two wrongs don't make a right. All I can say is that we must support OpenAI with all our energies, since they have the proven track record of innovation here, not some substandard and cheap product like DeepSeek...

DeepSeek is open-sourced, so even if the data is stolen, it's being recycled back into the public domain and publicly available for free. There is no going back, open source LLMs is the future.

You can't say the same about ClosedAI's stolen data, they enrich themselves and charge a higher price on it.

Any criticism ignoring the open-source aspect of debate is essentially a joke - nobody will take you seriously.