OpenAI Furious DeepSeek Might Have Stolen All the Data OpenAI Stole From Us

Hahahahaha

"The Chinese came, and they stole all our gubbins!"

"We must tell the media! they will help us!"

OpenAI.... probably.

Based coverage from 404.

Well I always thought I knew what the smallest violin looks like!
Look at its widdle toes! 🥹
- Those are clearly fiddle toes, in this case.
That's lovely.

Would you download a LLM?

Would you hug a face?
Yes, but I don't have Nvidia hardware to run them
I would. I gave llamafile a try, and while it's good, they have an issue with parsing responses that have {{ and }} in them, so it's kind of useless for Laravel development unfortunately lol

Welcome, new industry heads. That's how it works. China takes a car, picks it apart and builds a cheaper car. That's what they've been doing for decades now.

That's par for the course, but it's hilarious that openai "we have to get copyrighted material for free because fuck you" is pulling that defense now.
- Yep.
We got angry when Japan did this in the 60s and 70s. I'm going to paste part of the opening from Neal Stephenson's "Snow Crash."

Why is the Deliverator so equipped? Because people rely on him. He is a roll model. This is America. People do whatever the fuck they feel like doing, you got a problem with that? Because they have a right to. And because they have guns and no one can fucking stop them. As a result, this country has one of the worst economies in the world. When it gets down to it -- talking trade balances here -- once we've brain-drained all our technology into other countries, once things have evened out, they 're making cars in Bolivia and microwave ovens in Tadzhikistan and selling them here -- once our edge in natural resources has been made irrelevant by giant Hong Kong ships and dirigibles that can ship North Dakota all the way to New Zealand for a nickel -- once the Invisible Hand has taken all those historical inequities and smeared them out into a broad global layer of what a Pakistani brickmaker would consider to be prosperity -- y'know what? There's only four things we do better than anyone else

music

movies

microcode (software)

high-speed pizza delivery

The Deliverator used to make software. Still does, sometimes. But if life were a mellow elementary school run by well-meaning education Ph.D.s, the Deliverator ' s report card would say: "Hiro is so bright and creative but needs to work harder on his cooperation skills."
I'm kinda reminded of the tale of how the Zilog Z80 processor chip had dozens of little "tricks" built into it. It was being produced in Japan which at the time was famous for their chip production and for copying chip designs. Apparently their little tricks were baffling enough that it delayed the appearance of knock-offs chips by half a year.
Ah the Burger King model
looks at all my non-critical electronics.....

enshitification smells like Chineseium

That said, I like cheap non-critical crap
- I bought my parachute from AliExpress and my reserve chute from Temu.
With zero investment in innovation. They just wait and steal the work. Easy to undercut American companies when you have no R&D costs.
- Forgot that they are only able to do so because the Western capitalists have been trying to snuff out domestic organized labor and thus dumping huge sums of money into Chinese production for half a century.
- I guess, but very often private innovation builds upon a bunch of fundamental research funded by the tax payer. Then the private sector patents it, and brings it to market, overcharges and earns billions. Tough luck if China gets better at this game.
- If it was just pure copying the best you could hope for is that you match the performance of your competitor. To exceed their performance genuine investment must be made.
- And since China is not a party to any Western IP trade agreements and not bound under international trade law, the only solution is a) diplomacy or b) war

There’s substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI’s models [...]

I will explain what this means in a moment, but first: Hahahahahahahahahahahahahahahaha hahahhahahahahahahahahahahaha.

LMFAOOO 😂😂

It sounds like they just followed the precedent set by American companies. Maybe don't steal data and your data won't get stolen?

So what

Gonna cry, technofeudalist lord?

1000045471

Is there evidence that DeepSeek is an OpenAI distillate other than OpenAI and Co's protestations?

It's literally impossible. I tried to explain it here: https://lemmy.world/comment/14763233

But the short version is OpenAI doesn't even offer access to the data you need for a "distillation," as the term is used in the LLM community.

Of course there's some OpenAI data in the base model, but that's partially because it's splattered all over the internet now.
- Thank you 🙏
Not distillate, they just trained on the outputs of openai

I don't particularly agree that LLMs are nothing more than spicy autocorrect nor is it the end all be all super tool that AI tech bros overhype to death. It's somewhere in the middle with good usefulness and utility like any other tool as long as it's used properly. Like any other tool.

That being said, even if it was just a spicy autocorrect, it's the best damn autocorrect I've seen in my life and won't screw around with my ducking cuss words like the dumb autocorrect systems
- I'm pretty sure the autocorrect ducking with fowl language is because it has those words specifically marked not to come up in autocorrect, because kids do also use phones too and that would be enough to cause enough of a stink that no company wants to deal with it.
- I think it should be akin to the market effect of PDA’s in the 90’s.

Ironic and hypocritical

a) to expect China not to steal every piece of design they can lay their hands on is foolish and should be part of every tech companies contingency planning, and investor consideration b) given that deepseek seems to have condensed the processing, i can only imagine openai can now use their processes to make the high end chips work just that much more efficiently

To expect any company to not steal is foolish. All development builds on previously proven ideas.
- Property is theft, after all.
Yeah, but so can others, their vaunted "we trained for years for billions" moat is gone

Competition is back on the menu, and US VCs don't like to play that game
- It's their least favorite game right behind Q&A at the SEC.

Does the OpenAI TOS even cover no-take-backsies ?

Stealing for me, not for thee or Li
Finders keepers giveth, finders keepers taketh away

Oh the irony 😂😂😂😂😂😂😂

Good, I hope this is how the AI industry dies.

How would this cause it to die?
- If something ceases to be profitable, it gets no attention from corporations.
  
  Even something as simple as Deepseek replacing subscription services would tank these corporations who are banking on those fees.
- It would kill off American ai industry because it’s a pyramid scheme of bullshit and money laundering
- They think gpt is ai....lol

BAHAHAHAHAHAHAH 😂

So that’s what they did with the data from TikTok

Man that would be as dumb as using Reddit or 4chan to train models.
- Now wait just a minute...
- Unless they just used Chinese TikTok, where the government enforces algorithmic preference for educational content.
  
  Unless, maybe, that's why it's better?

"Furious"

Somehow I don't think that word applies to companies

Tell that to nintendo, whenever someone does something legal.
Hey, companies are people too, as has been proven in a court of law
- I'll believe it when Texas executes one

Wow, look at the size of the tear this brings to my eye.

Wherever you fall on the anti-AI spectrum, I thought after the past 2 decades of piracy we had come to the conclusion that you can't "steal" data, copying != Stealing

If anything, this is kind of making people realize the opposite. It isn't stealing when it is corporate (or creator...) but it is TOTALLY stealing when it is individual people... who aren't authors or artists.

The fun part is that "creating" datasets for training steal from everyone equally.
- And when it comes to authors and artists, it amounts to wage theft. When a company hires an artist to make an ad, the artist gets paid to make it. If you then take that ad, you're not taking money from the worker - they already got paid for the work that they did. Even if you take a piece from the social media of an independent artist and make a meme out of it or something, so long as people can find that artist, it can lead to people hiring them. But if you chop it up and mash it into a data set, you're taking their work for profit or to avoid paying them for their skills and expertise to create something new. AI can not exist without a constant stream of human art to devour, yet nobody thinks the work to produce that art is worth paying for. It's doing a corporation to avoid paying the working class what their skills are worth.
- Nuance not your strong suit eh?
I can’t recall a time when I downloaded an album, pretended I made it, and tried to sell access to it.
- This would be more like buying a van halen album, learning eddies style, writing and recording my own van halen style songs and selling those. Still an infringement?
- That's not how training works with LLMs at all
Uh, who is "we"? Piracy is still illegal and not everyone approves of it.
- Maybe in the mainstream, but the open source, socialist and anarchist communities that populate lemmy tend to be very critical of ideas like intellectual property and copy right.
- Oh, you're one of those weirdos that report people shoplifting at Walmart and probably was also the "Teacher, your forgot our homework" kids

I want to punch his face. (Proceeds to Wim Hof breathe)

Openai has been trying to switch for-profit. They must be loving the hype. Now they have an excuse.

OpenAI is open source. No shit, that's the point.

there's nothing open about OpenAI