130
submitted 2 months ago* (last edited 2 months ago) by plinky@hexbear.net to c/technology@hexbear.net
all 37 comments
sorted by: hot top controversial new old
[-] edge@hexbear.net 100 points 2 months ago

as concerns grow over a potential breach of intellectual property

Oh no, the plagiarism machine's intellectual property was plagiarized!

[-] edge@hexbear.net 74 points 2 months ago* (last edited 2 months ago)

The Gizmodo article on this is good.

OpenAI Claims DeepSeek Plagiarized Its Plagiarism Machine

OpenAI and Microsoft are big mad that Chinese AI startup DeepSeek has stolen their market share and, possibly, portions of their code. It’s a deeply funny claim from the company that made ChatGPT, a program it once admitted couldn’t exist without free access to all the copyrighted data in the world.

"...And they can kind of suck the knowledge out of the parent model."

Got it. Distillation is when one AI sucks off another AI. So it’s a fancy word for copying.

data-laughing

[-] plinky@hexbear.net 39 points 2 months ago

Me, distilling some twixes from chain store: ancap-good

[-] crime@hexbear.net 33 points 2 months ago

When I was studying machine learning a decade ago we talked a lot about how you can clone a model by training it on another model, it's well-known 101-level stuff. The way OpenAI is presenting it as something shocking is sooooooo disingenuous it's driving me crazy.

[-] marxisthayaca@hexbear.net 26 points 2 months ago

It's shifty when the foreigners do it.

[-] OptimusSubprime@hexbear.net 19 points 2 months ago

Got it. Distillation is when one AI sucks off another AI.

volcel-judge Motherfucker, not on my watch!

[-] AtmosphericRiversCuomo@hexbear.net 8 points 2 months ago

Uh oh, his tech is really starting to snowball.

[-] TraschcanOfIdeology@hexbear.net 14 points 2 months ago

Distillation is when one AI sucks off another AI

So that makes two Chinese machines that fulfil their purpose by sucking off. Deepseek and the one they use to extract semen.

(yes i was reminded of that machine by the latest TF episode, i'm not a very original person)

[-] john_browns_beard@hexbear.net 3 points 2 months ago

Got it. Distillation is when one AI sucks off another AI. So it’s a fancy word for copying.

I think there's something wrong with your thesaurus

[-] nohaybanda@hexbear.net 5 points 2 months ago

You say that like you've never invited someone over to "copy their homework"

[-] Speaker@hexbear.net 5 points 2 months ago

OH FUCK I'M GONNA PAAAASTE

[-] Cimbazarov@hexbear.net 60 points 2 months ago

Capitalist pretending to like competition. Dude has to justify why they should give him a gorillion dollars when you can give a small group in China access to their technology and produce better results with a fraction of the cost

[-] hotcouchguy@hexbear.net 37 points 2 months ago

I'm not mad, I'm invigorated actually

[-] NPa@hexbear.net 24 points 2 months ago

Trillion dollar market loss? Who care? 🤣🤣🤣

[-] Z_Poster365@hexbear.net 10 points 2 months ago

we will obviously deliver much better models than the inferior chinese

[-] Evilphd666@hexbear.net 9 points 2 months ago

Better grift fast Altman. Markets are crapping the bed and crapitall might be a bit tight when shit hits the fan.

[-] aebletrae@hexbear.net 56 points 2 months ago

‘Scraping whatever you want, permission or no, is good and necessary.’

‘No, not like that.’

[-] Barabas@hexbear.net 44 points 2 months ago* (last edited 2 months ago)

It is kind of hilarious how the Silicon Valley dweebs got their lunch eaten by people who treated LLMs as an LLM instead of thinking that they're going to create god.

Maybe this will get all the "We are creating Skynet" drama to die down a bit.

[-] Acute_Engles@hexbear.net 36 points 2 months ago

Damn that's crazy. Anyway it seems like in China they use different laws so don't care

[-] dkr567@hexbear.net 30 points 2 months ago* (last edited 2 months ago)

That's rich coming from a fraud that literally wouldn't be where he is without our data he and his cronies plagiarized to train his overpriced piece of shit program.

[-] barrbaric@hexbear.net 27 points 2 months ago

I feel like the entire point of naming it OpenAI was just to do propaganda to trick people into thinking it was open source or something when IIRIC it's been a privately held company all along. No idea how effective it's been.

[-] Enjoyer_of_Games@hexbear.net 3 points 2 months ago

The whole point of the open source movement is capitalist recuperation of the libre software movement.

Now they have built the plagiarism machine to launder open and libre code into closed source code and consider their position secure enough to finally drop the pretense of being "open".

Like the Nazis the name was never more than a means to confuse and lure people away from the genuinely liberatory movement and the betrayal inevitable.

[-] piggy@hexbear.net 25 points 2 months ago

You can distill your own models based on your own models, the fact that OpenAI isn't doing this is more evidence that they are "competing" via investment capital and not tech.

[-] JohnBrownsBussy2@hexbear.net 23 points 2 months ago

OpenAI does this all the time. For example, all the models they offer for free users (like o1-mini or 3.5-turbo) are distilled models. The DeepSeek R1 is impressive because it's not a distilled model in the technical sense, but still has the cost benefits of distillation.

[-] Z_Poster365@hexbear.net 6 points 2 months ago

right if it was so easy to "distill" and "copy" then why haven't they done this to themselves for a more efficient model?

[-] godlessworm@hexbear.net 23 points 2 months ago

ok so why does your shit suck if they stole all your ideas then bitchass? it doesn’t even add up

[-] Z_Poster365@hexbear.net 18 points 2 months ago

westoids always do this anytime news comes out about Chinese technological breakthroughs beyond what the west has discovered. They "stole it" even though they are the most advanced in the world.

[-] Tabitha@hexbear.net 20 points 2 months ago

I'd install a chrome/ff extension that sends all my scrapables to a Chinese AI trainer.

[-] Robert_Kennedy_Jr@hexbear.net 11 points 2 months ago

xi-plz take my data

[-] gay_king_prince_charles@hexbear.net 18 points 2 months ago

There is a reason so many tech people call it ClosedAI

[-] Pavlichenko_Fan_Club@hexbear.net 15 points 2 months ago

i-spil-my-jice nooo my business model!

[-] woodenghost@hexbear.net 11 points 2 months ago* (last edited 2 months ago)

We can't show you any evidence, but here is why you shouldn't stop investing in us and by the way all these graphic cards are totally necessary after all and whoever claims they don't need them obviously stole from us (still can't show any evidence).

[-] AtmosphericRiversCuomo@hexbear.net 11 points 2 months ago

Grok will tell you it's a product of both OpenAI and Anthropic when asked, but no one cared until Chyynah.

[-] Evilphd666@hexbear.net 8 points 2 months ago

The proprietary code

If then else and or nor print.

[-] AvocadoVapelung@hexbear.net 5 points 2 months ago

does this mean they're trying to get it taken down?

this post was submitted on 29 Jan 2025
130 points (97.8% liked)

technology

23630 readers
375 users here now

On the road to fully automated luxury gay space communism.

Spreading Linux propaganda since 2020

Rules:

founded 4 years ago
MODERATORS