China is leading open source
Posted by TheLogiqueViper@reddit | LocalLLaMA | View on Reddit | 292 comments
Posted by TheLogiqueViper@reddit | LocalLLaMA | View on Reddit | 292 comments
Enough_Court_6212@reddit
True 😁
Rili-Anne@reddit
Honestly what makes me happiest is that an open weight model can't be FUCKING STEALTH QUANTIZED INTO USELESSNESS AFTER IT PERFORMS WELL ON THE BENCHMARKS, GOOGLE
whatswimsbeneath@reddit
What did I miss here?
VonKyaella@reddit
REMEMBER WHAT THEY TOOK FROM US
LostMitosis@reddit
Now hollywood and CNN have work to do.
SpecialBeginning6430@reddit
Yeah we should allow China to control the world.
MiddleCelery6616@reddit
As opposed to let muricans do it?
SpecialBeginning6430@reddit
I trust a country which cant elect a dictator versus a country that's had two dictators
FormalAd7367@reddit
So, one of American AI companies will have a video of them using AGI but inadvertently lost all the record?
hendy0@reddit
interesting
agdnan@reddit
God bless China 🇨🇳
TimeTravelingBeaver@reddit
I LOVE DICTATORSHIPS ❤️🇷🇺🇨🇳🇮🇷🇰🇵❤️
Sudden-Lingonberry-8@reddit
you forgot signapur
abskvrm@reddit
There can be only one western friend dictator: Satanyahu or MBS.
TimeTravelingBeaver@reddit
Bukele
Fit_Flower_8982@reddit
An oppressive dictatorship working faster than anyone else to turn the black mirror into reality? No.
agdnan@reddit
Western propaganda has you in choke hold
nrkishere@reddit
He is correct tho. As much as I appreciate open weight models from Chinese companies, the government imposed surveillance and censorship sucks hell.
Tech_Romancer1@reddit
https://www.youtube.com/watch?v=LAEwoYIYgJA
ROOFisonFIRE_usa@reddit
Is that not what the United States is becoming? The only reason we're behind is because we're technologically behind as a society.
-
Even if the United States has better technology in secret. The Chinese are living better than we are day to day. Better trains, better roads, better telecom, better manufacturing, you name it, it's probably being done better in China or in the east today.
KazuyaProta@reddit
That's not how it works even according to China itself
Tech_Romancer1@reddit
How is it not? At the very least they don't have a housing crisis (at least not in terms of artificially limited supply, thus creating a false demand and criminal levels of rent seeking) and possess walkable cities/competent public transit.
If you're referring to the meme about the Chinese being overworked plebs in factories, its not like the US has them really beat in this regard. Most in the US make below the poverty line, live paycheck to paycheck due to rent, some work two jobs and still barely make ends meet. Not to mention the price gouging on groceries and the rampant and open disregard of the population in favor of oligarchs.
It can't be the government spying on you, since the US does that as well and arguably far more insidiously.
So what are you referring to?
PidgeyPower@reddit
You've either never been to China, or more likely, never been to the US. The fact you have "usa" at the end of your username gives away exactly what you are, a phony imposter. It's pathetic what China has to do.
ROOFisonFIRE_usa@reddit
I've never been to China, but I doubt the articles and recounts of their trains, manufacturing prowess, or population count are made up.
-
The bottom line is that we are a fraction of their population, yet they are managing more efficiently. If we had as many citizens in the United States it would be chaos. We're barely managing our current population effectively. Just look to the university students to know who's ahead...
PidgeyPower@reddit
In your other replies you keep calling the US "they."
Premier Li Keqiang in 2020 said that over 600 million people in China live on less than 140 dollars a month. How efficient is life for those 600 million?
Also, sure, let's look at the university students. What direction are they headed? To the US. How many American students are headed in the opposite direction? Barely any. What direction is the money moving? Are the billionaires of the US still investing in China? No, they're trying to get their money out. Yet Chinese billionaires are money laundering everything they have out of China to avoid sharing their wealth. What is it that you're even trying to argue here? Your points don't stand to reason.
ROOFisonFIRE_usa@reddit
We can't really talk about wages when the cost of living and social services is organized differently https://youtu.be/eoIFOJDSkCQ?t=509. The video is of a western women talking about her experiences in China and how much her rent was in the city vs in a more rural area... $550 in the city, $370 in the more rural area... My rent is 1400 and I'm not getting a whole place to myself for that... Total rent is ~2200-2500 in my area!! In addition they have things like Universal healthcare in China meanwhile... I don't have healthcare at all and even when I'm paying for it, it hardly covers anything and I pay out of pocket until I've put in 4k out of pocket.
-
I'm not here to abdicate or say China is perfect. I know they have alot of issues as well and I wouldn't be interested in moving there myself, but my main point was that things aren't so much better in the United States and if our population was the same things would probably be much closer to how they are in China.
-
The University students come here and then often are going back home or to other places. The money is starting to leave that much is clear... see the decline in bond market purchases from foreign nations or the austerity measures the United States is taking because we can't afford even our current level of public services. China is playing the long game... Thats why so many nations are participating in the belt and road initiative and why so little investment and partnerships from the United States are taking place in Africa and South America. They are building damns, roads, internet, and trains all over the globe. Not US.
-
My main point is United States and China are not vastly different from Americans experiences just framed differently. At the end of the day citizens from both countries are under a great amount of surveillance that is only going to grow as population and potential threats increase. Thats regardless of what China is or isn't doing. Palantir, flock, ring, oracle, and the likes aren't going anywhere and have intentions to catalogue all of us.
PidgeyPower@reddit
I tell you the Chinese Premier in 2020 said that 600 million people in China (roughly half the population) lived on less than 140 dollars a month. You say rent can be as low as 370 in rural areas as a counterpoint. Please tell me you're just a bot and you didn't actually think this was a valid response. Either way, I'm probably going to leave this here. But this is the quote in case you are real, and it might help you going forward:
"The average per-capita annual income in China is 30,000 yuan (USD 4,193), but there are over 600 million people whose monthly income is barely 1,000 yuan (USD 140), not enough to rent a room in the Chinese cities"
ROOFisonFIRE_usa@reddit
I've never been to China.. I'm just speculating based on what I'm reading. I think what you are saying may be true, but in what context I'm not sure if I can even comprehend. Not because I doubt you, but because I don't have enough information to postulate on if that is a reasonable amount of money or an amount that would cause suffering. You can really feel the disparity between how far a dollar goes in the the West vs how far it goes in many third world countries.... It's not really a good comparison...
-
I guess I'm just going to have to visit once so I can stop talking out of my ass and see how their living in person..
-
I'm not denying what you are saying, I just don't really know what that means and I'm curious. I saw a video of 70 year old people carrying heavy ass food to market and I was like damn it's about time to retire right guys??? I'm not here to champion China, more a mild curiosity based on circumstances at play.
-
I am however displeased with how things are being run in the United States and think we've really sold our soul lately. Nothing happening right now is patriotic or maintaining freedom. The Taliban won.
SpecialBeginning6430@reddit
Nice whataboutism.
Xi Jinping has been in office longer than any US president and he will be there until he dies or loses a power struggle.
ROOFisonFIRE_usa@reddit
I'm not here to defend China really. China bad too. My point is that the United States isn't far behind in regards to a surveillance state. It's less to do with Trump or Xi and more to do with a natural response to increasing entropy.
Optimal_Effect1800@reddit
Still much better than mysantropic corporate dystopia on other side of ocean...
Minimum-Ad-2683@reddit
Corporate dystopia… I like that term
secunder73@reddit
It fits both China and USA lmao. Thats what needed to be a hegemony
YekytheGreat@reddit
I have come to uphold my promise! Let the sun shine upon this lord of cinder!
nrkishere@reddit
there are open source models from big tech as well. It is only Anthropic (which I doubt that considered "big tech") that is vehemently anti-open source.
Also Alibaba, Bytedance and Tencent are big tech themselves (and were vastly closed source until recently)
genshiryoku@reddit
Anthropic isn't anti-open source. They open source all of their alignment research and tools, like their recently released open circuit tracing tool which is very cool and useful.
It's their genuine conviction that open weights of (eventually) powerful models will result in catastrophic consequences for humanity.
Their alignment team is the best in the industry by far and I respect their work a lot.
tengo_harambe@reddit
The only catastrophic consequences they are worried about are to their bottom line lol
Pedalnomica@reddit
Lots of the people at Anthropic have been concerned about human extinction level consequences since before they had a bottom line.
You may disagree with their decision not to open source models, (like I can't imagine releasing Claude 2 would matter at all now) but I'm pretty sure they have serious concerns in addition to their bottom line.
llmentry@reddit
And yet ... they're still there, happily churning out the models. If you really thought that LLMs are an extinction-level threat, and you stay working in the industry making the models ever more capable, what does that say about you?
I could be wrong, but I strongly suspect that a lot of Anthropic's "OMG Claude is so powerful it's the End of Days if its weights are ever released to the wild!" shtick is marketing fluff. Our model is the most dangerous, ergo it's the smartest and the best, etc. It's a neat trick.
ColorlessCrowfeet@reddit
They're trying to lead in what they think is the right direction, but with a lot of real-world constraints. Steering in a different direction is leadership. Standing still isn't. That means churning out models.
randylush@reddit
It’s just ridiculous marketing
ColorlessCrowfeet@reddit
Nope. It's best efforts motivated by deep fears and hopes.
And colored by marketing.
ninjasaid13@reddit
Their quack paper implying that LLMs are sentient or some crap like that in their conclusion makes me doubt that.
ColorlessCrowfeet@reddit
They're afraid that at some point AI systems can be hurt in a meaningful way. As are many people in philosophy departments who own no Anthropic stock. The more we know about the subtle and fluid processes in LLMs, the more we know how little we understand.
ninjasaid13@reddit
wtf is meaningful way?
ColorlessCrowfeet@reddit
The difference between hitting a rock with a hammer and hitting a cat.
ninjasaid13@reddit
So anthropic believes in magical thinking.
ColorlessCrowfeet@reddit
Acknowledging ignorance isn't religious faith, and discovery can uncover whole new realms of ignorance.
For example, increasing knowledge of neuroscience discovered complexity that hadn't been anticipated and wasn't understood. Same story here, and researchers often draw parallels.
ninjasaid13@reddit
The problem here is that you assume the complexity means greater capability, rather it just means that the complexity putting bounds of what you originally thought.
For example General Relativity put bounds on Newton's theories despite General Relativity being something more complex.
ColorlessCrowfeet@reddit
Yes, the fact that the growth of scientific knowledge can reveal constraints, rather than possibilities, is important and widely misunderstood (especially by sf authors!).
If we define "current AI systems" narrowly (autoregressive models with exclusively downward attention?) and also construe "sentience" narrowly, I'd be inclined to agree. Let's leave it at that. You might enjoy discussing this with Opus 4. It's damned smart and seems to love the subject!
BackgroundMeeting857@reddit
Wait I forget are anthropic ran by some sort god like being that can do no harm or some shit lol. They are human like the rest of us. Such a moronic take "Don't worry, the corpos will stop us from shooting ourselves, I promise"
ColorlessCrowfeet@reddit
What are you talking about?
dmitry_sfw@reddit
Orh, are we talking about ethics, safety and trust? Sure, let's talk.
Why don't we start by calling Anthropic's scam with "safety testing" where an LLM was "blackmailing an employee with revealing an affair to prevent them from turning it off" for what it is: evil and calculated lies.
Everyone involved in this scam in any way, and anyone remotely technical who knows how the LLMs work knows this to be a blatant manipulation. Yet it works like a charm and this hot garbage makes rounds in the mainstream news.
If they are capable of that, it's obvious to anyone with half a brain that these unethical people should be the last ones you should listen to when it comes to "safety" or "concerns".
And it's clear as day why they are doing that.
it's a clear ploy for regulatory capture. They want to have regulations where it's effectively illegal to work on LLMs for anyone else but themselves.
And thankfully it sounds a bit more far-fetched now, but let's not forget that just a couple years ago the plan was working great. Anthropic, OpenAI and friends spent billions to lobby the Biden's administration and got really far. There was an executive order and president's advisory committee with lots of folks from Anthropic.
The committee also included such experts on existential risks as the FedEx's CEO, in case someone has any doubt whether it was the real life Avengers protecting humanity or just another smoozing fest of powerful insiders.
In an even sadder timeline than this one, the one where there was no Deepseek or even Llama, the one where the new administration was in on this grift as well, they already succeeded.
And by the way, If there was any justice or reason in our society, anyone responsible for this blatant scam would completely lose any credibility in the industry. They will be known as "Anthropic scammers" for the rest of their careers.
But I am not holding my breath.
ColorlessCrowfeet@reddit
Not everyone is lying scum. If you pay attention to what Anthropic's leadership has been saying and doing since they were founded -- and in many cases, for years before that -- it's clear they aren't scammers. They walk the walk, even if you think it's the wrong direction.
xchino@reddit
Plenty of con artist are true believers in their own grift.
ColorlessCrowfeet@reddit
And plenty of people will be in denial about what AI will be able to do until very late in the game.
xchino@reddit
That cuts both ways and proves how inane legislation by speculation is.
Pedalnomica@reddit
You can think think something is bad, that doesn't make it a grift.
Like it's pretty clear what their business model is, and I didn't see how it's a grift. Like sure they may go bust and leave the VCs hanging, but I don't think they are hiding that.
Affectionate-Hat-536@reddit
You can force people to do open source by shaming and call out for being commercial. Being open is in spirit and nature. Some would have a business model other than open source model and hence are ready to do it. Why do you want people to spend millions on and then give away freely. This will have no winners and a race to bottom and finally the last standing will basically not open source and create monopoly. We need both sides to innovate.
tengo_harambe@reddit
Eh there's nothing wrong with not open sourcing their moneymaker but the fake righteousness is cringey.
I don't see this company surviving in the long term anyhow. Google will eat its lunch.
zdy132@reddit
I don't see any "indie" LLM company surviving google in the long run, now that Google seems to have found a grip.
Google just has access to so much data. The only companies that can compete long term are other similar giants, like Meta, Bytedance, and maybe Twitter(X) with their xAI.
ambassadortim@reddit
Idk if Google found a grip. They may have started sharing their tech strategically to protect their business income in search, advertising, and other areas. This is why I think if they see less ad revenue, and decided to take AI market share, where will they use AI to make up lost search revenue? Ads on AI answers?
YourNonExistentGirl@reddit
Yes.
With their Research, Deep Research, Google Search tools integrated into Gemini.
SEO for AI search is so 2024.
Admirable-East3396@reddit
you watch 2 unskippable ads to get a less hallucinated response, if you buy premium they still show you ads but now they can be skipped and the quality of answers are objectively better.
OkTransportation568@reddit
It will become your bottom line ones AI takes your job. It’s inevitable, but they don’t have to contribute to speeding it up. I heard Anthropic attempts to stay at other line with other AIs without pushing it forward. It’s a legitimate concern, and many are already predicting the numbers of jobs being replaced by AI. None of the others are open sourcing models anywhere close to their flagship ones.
sassydodo@reddit
Claude is also the most humane sounding model as of now. Also the only one that got "it's not well done it's congratulations" meme about steak
Western_Objective209@reddit
Claude seems to maliciously break alignment more then other models I've tried. I was using Claude Code the other day and asked it to fix all failing tests in a project; it was struggling with one of them and instead of trying to find the root cause, it erased a complicated algorithm and replaced it with a hard-coded value that matches the test. This is the kind of thing that a bad contractor who is just trying to close a ticket so their boss won't yell at them does. The Claude models are the best coding agents at the moment, but I feel like I have to be a taskmaster watching their output like a hawk because they often try to sneak things in that feel lazy/maliciously compliant
218-69@reddit
Their alignment team are the biggest grifters in the space. I'd rather they kept their "research" to themselves if their conclusion is "ai bad wants to take over humanity"
It's honestly a joke if you take that seriously or think it's anything more than an attempt to strongarm for regulation and public perception for a moat of their own.
ReasonablePossum_@reddit
Oh sure, them cooperating wity fkin palantir is "good for humanity"
Niightstalker@reddit
MCP is from Anthropic as well and open source.
nrkishere@reddit
anti open-source in context of AI models. They partnered with Palantir and tried to lobby government to prevent open source LLMs
this is why they partnered with fucking Palantir which supply military software to Israel (for targeted killing of Gazans)?
Open source models resulting in catastrophic consequences is total bs. It is like saying studying nuclear physics is catastrophic because a scientist can develop nukes in his backyard. Knowledge alone is not sufficient to achieve anything meaningful in real life, you need physical resources for that which are strictly scrutinized by governments based on the material.
AI access and control being limited in hands of small group of people create an uneven society. AI has potential to positively impact people's lives, it should be openly accessible for all. If Anthropic don't want to release their models, it is their personal business decision. But they can stfu from lobbying government or partnering with war crime enablers like Palantir
JustAGuyWhoLikesAI@reddit
China is sadly still pretty closed source on the image model side of things. ByteDance's Seedream 3 model, which is in 2nd place behind GPT-4's image, is unfortunately closed. Hunyuan (tencent) also closed-sourced. their latest Hunyuan 2.0 image model despite openly releasing Hunyuan image 1.0 and Hunyuan video last year.
nrkishere@reddit
For diffusion models, Europe will save our ass, don't worry.
phorouser@reddit
The entirety of Europe can't produce one competitive model. Mistral is good for smaller models, but last I checked was outgunned by Qwen and even Gemma?
Europe is not going to produce a proper model anytime soon because of their stringent regulations. US, obviously, and China obviously. Only other country I know of that could possibly maybe perhaps produce a good model is India, but I really doubt since all their models are absolute trash. I would think they have the infrastrucre though.
Why haven't I heard of any competitve models from Korea/Japan/Singapore? Am I dumb? They should be competing as well as Europe.
nrkishere@reddit
I'm talking about diffusion models. StabilityAi and Blackforest labs have some of the best image generation models out there. Recent flux kontext (released 2 days ago) is arguably the best image editing model from my personal experience and it will have a open weight variant soon
For Japan and Korea, they lack manpower and have different focus for AI. Both Japan and Korea have huge industrial robotics industry and their AI research is focused on industrial installations, not generative AI. I have no idea about Singapore, having less population than smaller cities of China can be reason
procgen@reddit
The future is going to be in multimodality, like 4o's astonishing image gen.
Hoodfu@reddit
Absolutely. Flux has created a whole ecosystem that's amazing. Chroma which is based on it is coming up fast based on the Flux architecture and is now surpassing it in artist representation and prompt following.
porzione@reddit
S Korea https://huggingface.co/LGAI-EXAONE
These models are very competitive, but with the worst possible restrictive license.
HugoCortell@reddit
OpenAI is also very much against Open AI. They keep teasing about how they are going to release an open model, but they never will.
aristotleschild@reddit
Yep, their company name is actual Orwellian double-speak
YourNonExistentGirl@reddit
Open AI for Countries will be a huge fuckin' problem.
HugoCortell@reddit
Yeah, it would. Empowering the citizenry is the greatest threat to tyrants world wide.
YourNonExistentGirl@reddit
I'm sorry, what?
HugoCortell@reddit
Governments want full control over ML/AI stuff, keeping it off the hands of regular people while using it themselves for mass surveillance purposes.
YourNonExistentGirl@reddit
I meant OpenAI for Countries, not open-sourced AI for countries.
HugoCortell@reddit
That's also awful. Governments should not trust a closed-source proprietary model from a hostile government.
pigeon57434@reddit
bro its literally been like 2 months since they said they would release an open model and they only occasionally mention it every once and a while people here so impatient its honestly impressive it will come out man it takes a few months to make a model they said they're not gonna release a currently existing model they're making a new one from scratch that takes a while
cantthinkofausrnme@reddit
They've been teasing for years, not two months 😳 which may never happen. Hey, maybe it will, but I give them credit for their earlier releases.
TheLogiqueViper@reddit (OP)
Open source models from big tech are not their main models
nrkishere@reddit
doesn't matter. This is "local llama" and Gemma, phi, granite models are extremely useful for most tasks. 99% of this wouldn't be able to run a model of Gemini 2.5's size anyway
LukeDaTastyBoi@reddit
Many (like me) can't even run DeepSeek locally. Still love it tho and hope they keep up the good work!
gripntear@reddit
So long as the DeepSeek team continues pumping out LLMs with little to no guardrails (It's just fucking text, you silly Americans. Get over yourselves), release Open Weights to the public (with permissive licenses), and be the cheapest solution out there for API use? Yeah, they're doing good (for now).
Admirable-East3396@reddit
chinese open source also arent handicapping the models by claiming "catastrophe for humanity"
BusRevolutionary9893@reddit
Chinese companies also aren't handicapped by our oppressive intellectual property law. Does the NY Times really own the knowledge they disseminate? I only have to pay the price of their newspaper to train my brain on its content. Why should it cost more for an LLM?
read_ing@reddit
You are not paying because NYT owns the knowledge. You are paying for the convenience of someone else gathering and presenting that knowledge to you, on a platter. Aka reporters, editors, etc, that’s who you are paying for and that’s why LLMs should pay for it too, every time they disseminate any part that knowledge.
BusRevolutionary9893@reddit
I could quote a New York Times article in another newspaper or television show and profit off it. It's called fair use. LLMs should be able to do the same and that's why they shouldn't have to pay more to do so.
BlurredSight@reddit
No way in hell this isn’t a bot funded by one of the big companies to change opinions on illegal data scraping
BusRevolutionary9893@reddit
Data scraping isn't illegal. At worst it's against a site's terms of service. However, I was never talking about data scraping. I was talking about copyright.
BlurredSight@reddit
Data Scraping news articles which are considered IP and protected by copyright and paywalled content is illegal
BusRevolutionary9893@reddit
Only if you sell or give it to someone else. If you have authorized access to something you can copy it.
Uwwuwuwuwuwuwuwuw@reddit
What are you even talking about? If LLMs had eyeballs and thumbs they could just read the newspaper like everyone else. They’re paying more for the way they’re accessing it, and the NYT is charging what the market will pay.
BusRevolutionary9893@reddit
And if a company training an LLM chose to access it like any normal person and used it as training data, it would be no different than than a news station using the same information to quote them in a broadcast they were profiting from. The courts will most likely, or should, come to the same conclusion. That will of course cost millions to litigate. Meanwhile China is kicking our ass because they don't have such absurd copyright laws. Intellectual property laws should focus on patents, that expire, not copyright. Should someone really be able to own something like the happy birthday song? Someone did in the United States for over 90 years.
read_ing@reddit
To access it like a normal person they would have to have a subscription to NYT. So, what’s fair would be that the company purchases a NYT subscription for each of their 100s of millions of users. I am confident that NYT would have no problem with that.
BusRevolutionary9893@reddit
Does a news station that quotes the New York Times have to have a subscription to the NYT for everyone of their viewers?
read_ing@reddit
They don’t need to because they have a financial arrangement instead thru contracts in various forms. LLM companies are welcome to do the same.
accountnumber009@reddit
You can't just make shit up and think people will believe you. The copy editor for a competing newspaper has 1 NYT subscription for the entire office to see what stories they are publishing and making their own. Happens every single day and has been happening even before subscriptions.
read_ing@reddit
Do read up on NYT wire services and syndication licensing. Here is a good place to start:
https://nytlicensing.com/content-media-organizations/
accountnumber009@reddit
Yeah man, they're totally paying licensing fees for articles that come out daily instead of putting it in chatgpt from a $1 account and getting a new article for themselves. Totally.
Just because they have a licensing company doesn't mean it's not a joke statement.
read_ing@reddit
https://www.cnbc.com/2025/05/29/amazon-ai-new-york-times-alexa.html
BusRevolutionary9893@reddit
No they don't. It's called fair use. Anyone can quote the New York Times or anyone or anything else for that matter.
thebrainpal@reddit
Bro is NOT an IP lawyer 🤣
__JockY__@reddit
Wholesale copying of data is not “fair use”.
BusRevolutionary9893@reddit
Training an LLM is not copying.
read_ing@reddit
Your assertions suggest that you don’t understand how LLMs work.
Let me simplify - LLMs memorize data and context for subsequent recall when provided similar context through user prompt, that’s copying.
BusRevolutionary9893@reddit
They do not memorize. You should not be explaining LLMs to anyone.
read_ing@reddit
That they do memorize has been well known since early days of LLMs. For example:
https://arxiv.org/pdf/2311.17035
There’s lot more research available on this topic, just search if you want to get up to speed.
__JockY__@reddit
I’m well aware of how they work, thank you. The issue isn’t that the LLMs are “simply” weights derived from the data (and more besides) in question, nor that the original information is or is not “retained” in the LLM.
It is the use of other people’s data at this scale that isn’t fair. Their data (which cost them a lot of money to create and curate) was used en masse to derive new commercial products without so much as attribution, let alone compensation.
It says “your work is of no value” while creating billions in AI product value from the work! This is not fair. It is not fair use, and retention of the original data is irrelevant in this regard.
read_ing@reddit
Do check who I responded to. But the rest of the point you made, is valid.
qroshan@reddit
just like someone with a didactic memory
read_ing@reddit
https://en.wikipedia.org/wiki/Eidetic_memory
qroshan@reddit
Thanks for the correction
read_ing@reddit
You are welcome. It was also the easiest way to point out eidetic is transient at best, in a small number of children and true photographic memory doesn’t exist.
ii-___-ii@reddit
but gathering a dataset probably is
BusRevolutionary9893@reddit
You can make a copy of something you purchased. You just can't sell it. I could use that copy, we'll say a video, and take a clip of it, video myself discussing it, and sell that video.
ii-___-ii@reddit
Sure, you can reuse limited pieces for commentary or quotes under fair use, but you can’t, for instance, record every video on Netflix and use that to make a commercial product, just because you have a Netflix subscription.
314kabinet@reddit
If the resulting commercial product does not contain copies of the copyrighted material then yes you can.
__JockY__@reddit
Not if it violates the terms you agreed to when you signed up for the service.
__JockY__@reddit
Obviously they had to copy the data to train the LLM, but I didn’t say copying. I said using.
The entirety of the hard-earned data and content was used by LLM trainers to create billions of dollars in value without so much as acknowledging the source of the data.
The LLMs could not have been built to their current standard without the data and content.
Therefore use of the data extends beyond fair and into commercial use.
It’s not fair use. It’s commercial use.
BusRevolutionary9893@reddit
You must be an artist or some kind of copyright holder. I really think you should learn about the purpose and flexibility of fair use. It's about balancing property rights, innovation, and the public interests. The same idea is why we have public libraries. Copyright holders flipped out when they became a thing too.
https://en.m.wikipedia.org/wiki/Fair_use
From the article:
Our copyright law is absolutely stifling United States innovation in AI, which is of extreme importance. It's why companies in China took ideas from over her, ran with it, and are leaving us in the dust.
BinaryLoopInPlace@reddit
What a silly mindset. Do you pay the people who wrote elementary school textbooks every time you do 2+2 in your head?
read_ing@reddit
You don’t need to, because your parents already paid for your elementary school textbooks that taught you how to do 2+2 in your head.
Don’t know where you were going with the tree imagining analogy and its relevance in this context, so going to pass on it.
BinaryLoopInPlace@reddit
By saying you don't understand the comparison you're either being deliberately obtuse or you don't understand the meaning of your own wording. There's a difference between paying for something once, versus paying in perpetuity for everything even remotely related to knowing about said thing's existence in the future.
The tree analogy is a mockery of the exact same rent-seeking mentality but applied to image models. Seeing something and learning from having seen it is not theft, and you don't owe anyone anything when you create new texts and new images inspired by what you've read or seen before. This is something that should be inherently obvious.
But when one's income relies on not understanding the obvious... Judging by your account and your posts, you don't have any genuine understanding of machine learning. You're pushing the "LLMs just memorize" halfwit take in other comments, the take that has been so thoroughly debunked it isn't even worth responding to.
Your only interaction with this community as far as I can tell is to randomly come in to this specific thread and shill for NYT.
read_ing@reddit
If you use the NYT content in perpetuity, you need to pay for it in perpetuity.
Not inspired, memorized - read the paper again maybe.
Posting a long diatribe doesn’t make your point anymore valid.
Mickenfox@reddit
Because rewarding people who write good content is good.
BusRevolutionary9893@reddit
Creating better AI is far more important.
DeviantApeArt2@reddit
Lol, Chinese companies aren't handicapped by anything, including IP, data collection and ethical guidelines. Meta got into deep trouble for torrenting some books, Chinese companies don't have to worry about that, that's why they will win eventually. Only thing holding them back are limited GPUs or else it would be total domination.
StyMaar@reddit
LOL.
Meta aren't in “deep trouble” at all, you'd be in jail for a small fraction of what they did, they will get away with it.
StyMaar@reddit
As much as I hate the current copyreich laws, it makes no sense to say US companies are handicaped by them when they have been very vocal about how they have been violating them from beginning.
lqstuart@reddit
lol yes China isn’t capped by any intellectual property law at all, almost as though they steal anything and everything they get their hands on
Extra-Whereas-9408@reddit
Yeah, you can do that. But collecting all of those articles, compressing them and selling them stinks.
shouryannikam@reddit
Nice try Sam Altman
AvidCyclist250@reddit
not half as bad gemini which rollercoasters strictly along guardrails and two-sides everything
Tchallaxxx@reddit
I'm confused what you mean. This thread is about China leading the open source LLM arena, and that is precisely taking in the role of a form of good guy in this situation...???
AvidCyclist250@reddit
It's their tried and tested strategy of destroying the competition. Happened with our solar industry here in Germany for example.
Tchallaxxx@reddit
We should do the same thing in industries our countries have a comparative advantage in. The consumers win with lower prices, and their sales at a loss can't go indefinitely. Only profit hungry monopolies would lose.
thrownawaymane@reddit
Also happened with rare earth metals in the US. It's an old strategy.
BinaryLoopInPlace@reddit
Huh. Gemini agrees with everything I say then gasses me up for being so smart.
it must be because I'm perfect.
218-69@reddit
Gemini is the best neutral model, from every model available. If anything, it has been a running joke that it will try to argue with you about most things that make people frustrated from being used to only soy models that agree with everything they say
CuTe_M0nitor@reddit
Yeah they distilled OpenAI model
SkyMarshal@reddit
They're just censoring things from them like Tienanmen Square and anything else the CCP doesn't like.
Admirable-East3396@reddit
am pretty sure thats the api which is censored since its hosted in china and has to follow certain laws, they arent making their open source shit purposefully on the name of danger to human kind, that censor is pretty minimal, you can have it output whole tienanmen square incident by just asking it to never use it in the sentence instead use code names it knows things and is actually pretty unbiased.
aristotleschild@reddit
What are these silly apologetics? Obviously they censor criticism of Xi and CCP just as they do in mainland China. Oh wait, Tencent owns north of 10% of this website doesn't it? LOL
KazuyaProta@reddit
That's a ridiculous definition or "uncensored
That's because the developers lack any means to stop a AI (essentially a large databank) from actually censoring those things without harming the output.
Admirable-East3396@reddit
what i meant is the api detects "tienanmen square" and removes it, the model hasnt gone through additional training to make it censored or biased like people were claiming on twitter when deepseek was released.
they can output nsfw with no problem while llama like models will not output it
yeah my definition of censored is wrong should have said biased. a lot of open source models by these big companies are made biased and censored
techno156@reddit
That's just on their end when hosting, which is fair enough. If you've the hardware to run it yourself, it will operated uncensored.
Alternative-Joke-836@reddit
I don't know. I haven't tried the latest and greatest but when it first came out I ran it locally. You could read its chain of thought talking about how it shouldn't talk about things that would be censored by the CCP. It would then say, can't talk about it as a response. Sure, I hacked at it and had it give me a better response but, to my understanding, it was just the model saying to itself that it shouldn't talk about it.
Anyways, never had the same issue with llama or other models out therr.
rz2000@reddit
Except, that’s not true at all for the open models themselves. You can believe people who are threatened by open models, or you can test the clains against the models themselves when hosted anywhere other than within China.
For example, using Kagi Assistant DeepSeek V3:
And, it goes on to describe government misdeeds and internal politics.
BoJackHorseMan53@reddit
It's a separate model for censorship. They don't bake the censorship into the models
Prudent_Elevator4685@reddit
Still quite uncensored
ErikThiart@reddit
It's the only way they can disrupt US tech and funding because they can't compete otherwise.
Minimum-Ad-2683@reddit
Why do people in the US always frame everything into a competition? I am genuinely amazed by “AI race”, “AI competition “ You would think they invented competing
ErikThiart@reddit
because that is exactly what it is. US economy is PVP
Minimum-Ad-2683@reddit
Problem is they think the rest of the world thinks that way
Affectionate_Use9936@reddit
If you’re implying that Chinese companies don’t literally steal company secrets and wipe out small businesses with import restrictions and dumping for the last 30 years, you’re delusional
ErikThiart@reddit
there is no civilization that comes remotely close. the US is on a absolutely insane scale.
can't blame them.
Minimum-Ad-2683@reddit
I couldn’t agree more
Lynx2447@reddit
Oh shut the fuck up. You have a post 8 months ago comparing germany to china. You're either a disingenuous hater, at best, or a bot.
DominusIniquitatis@reddit
Because "muh capitalism" can't cooperate and has to compete, probably? :)
DominusIniquitatis@reddit
Because "muh capitalism" can't cooperate and has to compete, probably? :)
sportoholic@reddit
Chinese Labs are a mean machine
sportoholic@reddit
Which Open Source Model I should use for transcribing Audio Calls? Calls are in Indian Languages. I have used Whisper Large v3 and v2 and they are not good enough.
Turbulent_Pin7635@reddit
I would invert the labels...
HeinrichTheWolf_17@reddit
Yeah, China is 100% going to allocate as much computation and electricity it can in order to compete with silicon valley corporations.
cfehunter@reddit
Realistically it's all getting nationalised sooner or later if it continues to develop. No government on earth is going to let a private entity have complete control of a hyper competent AI.
For the USA, China taking the lead would definitely be an incentive to do it sooner.
HeinrichTheWolf_17@reddit
I agree. It’s exactly why I don’t think the US is going to continue the Holy Roman Empire model for much longer. China can collectively pull all their resources together, the US is splitting all its own stuff up between different ‘Princes’.
I think eventually, Elon Musk, Peter Thiel, and Sam Altman are going to be the real losers here, because even though I believe that Silicon Valley higher-ups are trying to bribe out the Trump Administration so they can keep their walled off gardens, the government is definitely going to want to seize it once China passes Google.
cfehunter@reddit
Presumably they would pay them out, but if they're envisioning riding into the future at the head of an AI megacorp... it's not happening.
ThaisaGuilford@reddit
The white house are the good guys
Minimum-Ad-2683@reddit
Good to who though? Their bank accounts?
ThaisaGuilford@reddit
Israel
Neither-Phone-7264@reddit
? no one mentioned them.
ThaisaGuilford@reddit
Yeah I was just answering the question. I mean americans are the good guys, unlike russia and china.
PwanaZana@reddit
Chinese Labs release cool models that can run locally, but Tencent and Alibaba aren't the "small guy, fighting for the people". At all.
BlurredSight@reddit
I still do appreciate Alibaba being a free or at least cheaper provider for a lot of models on openrouter regardless if they have private interests with their more advanced models
Far_Car430@reddit
While I agree they are no small guys, but whoever releases open source model benefits the people. People can speculate their motivations, but it won’t change the good done to the community and I always appreciate that.
randomanoni@reddit
Community ... Sounds like communism! Liberty Prime is online!
PwanaZana@reddit
Yes, agreed. But when they think it is better for them to stop releasing open source/local, they'll stop. It's not because they are evil, they do need to make money off of these products, but they ain't good. :)
broose_the_moose@reddit
Exactly. And it’s only beneficial for them to release open source because they aren’t at the frontier. I guarantee you that if China was ahead of the US, their models would be closed.
scorpiove@reddit
Exactly, there strategy is the same as Meta's. Release open source to hurt your competition. But from their point of view they are doing it to hurt the US as a whole instead of Meta just trying to hurt OpenAI etc. I love open source but I recognize the devil behind the offering.
PwanaZana@reddit
Good example of that is Hunyuan 3D generation, it was open source, until they made 2.5, which is SOTA. Bam: that one is closed source, how surprising.
TheFuzzyFurry@reddit
You mean video generation? Or does it generate 3D models?
PwanaZana@reddit
The 3D model one. I haven'T tried hunyuan video generation, I've used Wan to a moderate amount of success.
serige@reddit
Isn't out government being ran by a bunch of xenophobic idiots that kick out the Chinese researchers so they ended up making these models back in China?
I_will_delete_myself@reddit
Only Alibaba has been fully open source. But it’s because they don’t need the money and it gets Americans using it so down the line they can build trust for future tools.
BidWestern1056@reddit
we not standing by in the US either
https://github.com/NPC-Worldwide/npcpy local models let us build amazing custom solutions for businesses that make SaaS economics a bit dicier. and plan w npc is to build and release fine tunes (https://huggingface.co/npc-worldwide ) and meta knowledge models and semantic methods that further reduce the need to ever use the large enterprise models to build gen ai apps.
sammcj@reddit
To be fair the Chinese labs are also producing closed source models but they make the weights and the inference code available openly.
Gold-Cucumber-2068@reddit
It seems like 99% of the people here don't understand that "Open Source" means you have the ability to recreate the binary blob. Virtually none of these models are truly open source. Open weights != Open Source. If you can't recreate it you don't know what the hell you're using.
Warm_Move122@reddit
100% agreed. I wonder if there is a list of open-source training codebases
sammcj@reddit
You have no idea how often I have to explain this to people that should know better.
segmentationsalt@reddit
It's hilarious that people think this is out of altruism and not just Commodotizing their complement
thrownawaymane@reddit
Fantastic article, especially since it has the perspective of being over 2 decades old. The concepts are evergreen.
Thanks for sharing.
Conscious_Cell_5194@reddit
Не много хотим.немного полутяти надо
uzzifx@reddit
China is currently leading every major technology in the world. Innovation is another level. Europe and US need to catch up.
Instrume@reddit
Small challenger needs more Chaos gods power, because if the CPC judges them worthy, congratulations! Instant Daemon Princehood for you!
memorial_mike@reddit
Acting like Chinese companies are doing this out of the goodness of their heart is laughable. Then there’s of course the fact that there are open weight models like Llama and Gemma. But that would compromise the openly pro-CCP narrative here.
cobbleplox@reddit
Propaganda much? Little guy china, what an excellent take.
abskvrm@reddit
With US corps acting like guardians protecting the precious hardware, starting trade wars and with their close source models, obviously China is the Main Character here. The David.
StoneCypher@reddit
You're delusional. Everything being discussed here is code given away by American megacorps.
abskvrm@reddit
"Code given by american corps" yeah they doing it for charity. lol
StoneCypher@reddit
Nobody said anything about charity
abskvrm@reddit
Beating US tech at its own game. Hell China W China. More EVs and LLMs baby!
StoneCypher@reddit
Whatever you have to tell yourself, champ.
Conscious_Nobody9571@reddit
Some people are delusional...
Lynx2447@reddit
This isnt a david vs goliath. It's a goliath vs goliath fighting while crushing a bunch of average joes.
Monkey_1505@reddit
ROOFisonFIRE_usa@reddit
Do you think this isn't a moon race?
Minimum-Ad-2683@reddit
I personally think that it is nonsense when every new innovation that should benefit society is turned into some “race”, and people just buy into it. But you know who benefits most from this BS the big corpos.
Imperator_Basileus@reddit
A very western centric viewpoint. What ‘corporations’ benefitted from the space race (not moon race) in the USSR and the Communist Bloc? The ones that did were ordinary people - scientists, engineers, planners, the Academies of Science, pilots, cosmonauts, et al.
slashrshot@reddit
it didnt start off as a race.
it is now.
both superpowers has seen the potential of LLMs
its not about winning, its about not losing.
Minimum-Ad-2683@reddit
What does that’s even mean man, the latest gemini models can’t even write a proper unit test for 30 percent of my backend service and now we are talking about superpowers and not losing?
ROOFisonFIRE_usa@reddit
There are certain new innovations where I agree the hype is too much, but this isn't one of them.
AI, space, new weapons, new telecom, new fab, new agritech. Those are things that can change the world and upset the apple cart.
Monkey_1505@reddit
Sure, it's kind of symbolic pageantry in the same way, at least for the US. They seem very intent on having the best chatbob.
fernando782@reddit
I will take it
nikilpatel94@reddit
OpEn SoUrCe
Quiet-Chocolate6407@reddit
OpenAI's name is pure irony, please actually be open!
No-Break-7922@reddit
Mistral, Meta, Microsoft in part, and Google in part are doing well as the Western open-source contributors, we're lucky to have them. Meta needs to pull themselves together though, they're a bit behind. OpenAI and Anthropic are lost causes and they'll soon be wiped out because of their greed. I hope to see Microsoft and Google go all in on open-source just like Alibaba.
Microsoft still hopes to be the monopoly of language models upon which the new world will be built. Their CEO recently said something along the lines of "SaaS either will build on our models or die", gave me a good laugh. Between the two, Google seems to be the one that's closer to the right open-source idea, which is maybe why they're ahead of Microsoft.
These two should finally understand that the future of inferencing is local, open-source, and lower parameter. Don't we see how fast the era of "wow a new model with MORE parameters" came to an end? That was the low hanging fruit of language model marketing, but China broke that wheel pretty fast with Deepseek and Qwen. I wanna see Google and Microsoft stop dreaming of becoming LLM monopolies and just do it (TM).
TrekkiMonstr@reddit
Meta/LLaMA?
i_am_m30w@reddit
Still waiting on you guys to figure out that China intentionally released open source so the worst case scenarios could unfold everywhere.
If only every country had strict internet censorship and draconian laws like them to keep them safe. /s
Worth_Geologist4643@reddit
The world is somehow becoming subscription-greedy. Ten fifteen years back cracks and some trojan horses saved the world. Now it's like subscribe to breathe.
Ylsid@reddit
I don't like it but the US is hardly stepping up to challenge
LelyaTwilightShifter@reddit
And those Chinese labs would be the same way if they could, it's just cheaper and easier to let the community do some of their work of course.
Don't be dense, there are neither good nor bad guys, just degrees of self interest based on the current calculus that will eventually shift as is the way of things
TinyNS@reddit
Or people could stop being selfish and collab to make an ultimate cohesive model.
I fucking hate humans.
bones10145@reddit
Haha China is definitely not the little guy
bones10145@reddit
Lol "Chinese open source". Nice pipe dream.
MerePotato@reddit
Ah yes, tiny underdogs with unlimited state funding
Radiant_Dog1937@reddit
But the news says they are on the verge of the collapse all the time. 🤔
abskvrm@reddit
Good for China Good for Us. Not US.
GhostArchitect01@reddit
Enjoy your backdoors
abskvrm@reddit
You'd appreciate Snowden.
Then_Advisor886@reddit
Guys is this Morgoth fighting legendary elf (forget his name)?
JerryWong048@reddit
Is this the promised communism?
raysar@reddit
China is not at all altruist. Aim is to destroy AI economy. It's an strategy. It'z cool for science and consumers, but not the bezt for future. China aim to be better than USA and make money with that.
For now it's cool and accelerate a lot AI ecosystem.
Ambitious_Toe_4357@reddit
What is the most popular operating system and remote source repository? Linux and Git, in my opinion. Guess what? Take a look at who is eating who's lunch in that arena. Torvalds seems to know what he was doing.
Plane_Ad9568@reddit
Unfair to meta
PersonalityNo4679@reddit
Godspeed
Holly_Shiits@reddit
Didn't really expect china to be a 'freedom fighter'
markeus101@reddit
I don’t think china is freedom fighting here lol. They know if they didn’t do something to break the western monopoly on AI the amount of money they would extract from the world and would shot up so fast that and they would never be able to catch up.
So they open sourced and broke the monopoly. Us getting free open sourced powerful models was just a byproduct not their intention.
Feeling-Buy12@reddit
So if China does it is because they are scared, and not because they want to help open source? you are very much biased and ate the western propaganda. China isn’t a block
slashrshot@reddit
china is a bloc.
jack ma disappearance and subsequent irrelevancy has proven that.
you can climb up.. to a point, anything above is on the auspices of the ccp.
Feeling-Buy12@reddit
That's your western view of this. China has their own view and it's acceptable. It isn't worst or better than us is just a different way of thinking. I think is wrong to look at china like it's a virus. They are a great society like any other, they aren't the best but they do their things good. They've given to open source and pushed those big companies to do it a little better. For me that's a major win. USA invest a shit ton of money on private companies and the citizen don't get jack shit at least in here we got deep seek
slashrshot@reddit
its not a western view. its a factual view.
u do not expand without the blessings of the ccp.
for example, all companies operating within china must give any data within china to the government when asked.
Feeling-Buy12@reddit
Same in usa and in any other country. we’ve seen how Trump bullied multiple companies if they don’t implement what he says, I thinj that’s quite the same.
slashrshot@reddit
no its definitely not.
for example: https://en.wikipedia.org/wiki/Apple%E2%80%93FBI_encryption_dispute
you do not or object challenge the Chinese government in court. you are just whisked away in the middle of the night.
Feeling-Buy12@reddit
You mean the case where the got the data? Let’s act like Apple didn’t help….
slashrshot@reddit
the point being is that there is no legal recourse in china to challenge government orders.
whatever the government demands is legal, that makes it a bloc.
not the case in almost all democratic countries like US or the EU.
markeus101@reddit
I am not saying china is a block and btw i love deepseek, i love qwen 3 and I’m very grateful for china to open source this tech but lets be honest here in any corpos they do absolutely nothing for the good of people they care about their bottom line mostly but china just showed us if you cant win the the competition you can certainly knee cap your opponents. And i think this trend will continue look at assemble ai they just open sourced chatterbox to take a chunk out of eleven labs which i am still laughing about. Since this strategy is only a win-win for us the consumers.
BlueArcherX@reddit
this post is pretty naive
Chogo82@reddit
Lots of people on here are okay with destroying American industries.
Open sourcing kills services industries. You don’t see China open sourcing their manufacturing technologies do you? It’s because they are actively protecting those and trying to hurt US services industries.
tenebrous_pangolin@reddit
Memes != reality
AtmosphereVirtual254@reddit
They're leading in open source experiments that have been validated at scale
dobablos@reddit
Another day, another dump of propaganda supporting China/CCP on Reddit.
Fearless_Weather_206@reddit
Lots of fanfare for Deepseek open source but why no American open source model with the same amount of love?
pigeon57434@reddit
because in the AI community for some reason people really love the underdog even if they're literally the biggest company on the planet if they used to act like the underdog they're cool this is the same reason why despite google literally being one of the biggest companies in existence that people typically don't like they're loved by the AI community because bard used to suck an going from sucking to amazing rapidly makes you look cool
TheLogiqueViper@reddit (OP)
China is pushing boundaries with their open source ai , deepseeks progress is so remarkable its competing with sota , lots of people are waiting for r2
Aydiagam@reddit
American company called OpenAI doesn't release models and advocate for closed source AI while chinese company is open sourced and released their biggest model right away. That's a satire
pigeon57434@reddit
openai do not "advocate" for closed source they're literally supposed to be releasing a open source model this june
S1lv3rC4t@reddit
China does what China can: Stealing
They even were to lazy to replace ChatGPT and OpenAI with their own names in the training data.
phree_radical@reddit
do that and you'd have a model with incorrect knowledge about ChatGPT and OpenAI
Sudden-Lingonberry-8@reddit
stealing from americans? BASED! keep going china
Monkey_1505@reddit
I honestly respect that more than desperately hiding the use of synthetic data, which we know for a fact every AI company is using (because they have to, there's no other way to generate reasoning data than using AI to make it), and generally trying to hide their processes.
Kinda based.
lmvg@reddit
Fun fact today is dragon boat festival in China, so that's not very nice of you lol
ElephantWithBlueEyes@reddit
Too much manifestation here. Like, people want to belong to that "open source" cool community and tell everybody about it but don't really contribute. But they want to seem cool so they tell about it on every corner
AppearanceHeavy6724@reddit
Keep in mind that one of the best 32b model, glm4 is made by Chinese government institution.
AppearanceHeavy6724@reddit
Keep in mind that one of the best 32b model, glm4 is made by Chinese government institution.
dodgepong@reddit
Spoken like someone who has never had a Chinese company take your GPL program and republish it with changes without distributing source code.
Fantaz1sta@reddit
Supply chain attack had entered the chat.
TimeTravelingBeaver@reddit
Which one for example?
Nonikwe@reddit
The correct use of the meme format! Love to see it...
Luston03@reddit
Pure Democratic Liberal Open(It's Open Because you can use it)Ai vs Brutal Communist Chinese Dictatorship Guess which one will make next greatest Ai Model?
RottenPingu1@reddit
How's the weather in Beijing, comrade?
cybran3@reddit
This should have the 3rd biggest guy, with smallest one being open weights models, bigger one closed models, and the biggest one being Deepseek.
Arcosim@reddit
Since their "small update" now beats Gemini 2.5 Pro in several benchs, that's accurate.
npquanh30402@reddit
It was just trained to beat the benchmarks, the benchmarks say nothing. Google is the biggest ad and data company, and their models are good in various tasks.
madaradess007@reddit
it beats nothing, you just quoting reddit bullshit
ihexx@reddit
it does win on AIME and LiveCodeBench; you can just look it up rather than quoting reddit bullshit
Any_Pressure4251@reddit
Look it up? why not try actually using it both can be for free.
Its just not as good as Gemini's.
randomrealname@reddit
CThis is factually not true, 5.5 mill was the final telraining phase, not the full r&d. I hate oai, they are closed, but the is an apples to oranges comparison.
Blarghnog@reddit
This is a lie. There is plenty of big tech open source AI in the west.
Primary-Editor-9288@reddit
China is doing to AI what it did with Manufacturing, took the idea from the west, improved it and made it way cheaper.
Maleficent_Age1577@reddit
Exactly. Orange man said Europe is unfair to them because we dont want to buy bad products from USA with high price.
Minute_Attempt3063@reddit
China, with all its flaws, is still pretty good with things like this
Accomplished_Ad9530@reddit
goddamn bot memes
TheTideRider@reddit
How ironic. We got OpenAI which is supposed to be open.