>They will probably never release the training data because that represents a large part of their competitive moat
Not that what you've written isn't the case. However, in addition to what you've written, (or probably even before what you've written), there's the fact that everything they're training on is stolen IP. Same with US LLM labs.
Let's not kid ourselve's about where the training data is coming from. They are not asking artists, writers, coders, content creators, etc etc etc for permission to use their creations.
Anthropic, Moonshot et al are doing incredible things, but we shouldn't gloss over the costs. Both present and future costs are kind of enormous.
Ok, but in 24 months that's the equivalent of downloading 100 variations of MSDOS. Torrents and vpn exist. I think this is more of a problem for corporations.
>but seem unlikely to surpass the companies that are training these models from scratch
Then why is it a problem?
Another serious question.
Trying to get my head around what the root of the objection is here. There must be some fear, but if that fear is not a fear of being surpassed in the market, then what is the fear?
A fear of competition causing a failure to recoup the trillions invested in AI via sky-high margins, and starting a (short) chain-reaction that causes the bubble to pop (or fizzle). A lot of people have a lot riding on the AI bubble not popping.
I didn't say it was a problem, I said it "mattered"
If I wanted to argue that it's a problem, I'd just say that companies investing billions in training frontier models should reap the rewards. And distillation is essentially theft.
First, whether we like it or not (generally not), a fuck ton of money has been invested in US AI companies, data centers, RLHF datasets amongst other datasets, etc. If that were to go to 0 that’d be quite disastrous. Alternatively, if it goes well, it’s great for the US (and to a lesser extent allies) economy and global standing.
Relatedly, tech has been a huge power house for the US economy for decades now. If the main driver of growth goes to China, what replaces it? Along with all the potential tax money, foreign investment, etc?
Next, patriotism / nationalism. This is very much a zero-sum game that defines who owns the future. Would you rather your country win or lose this? Lose this and you start losing talent, money, global standing, etc. That furthers a cascading effect that’s very negative. It also will likely create social strife with the knock on effects leading to even more bad populist ideas that just further diminish the country and tear society’s fabric apart.
> This is very much a zero-sum game that defines who owns the future. Would you rather your country win or lose this? Lose this and you start losing talent, money, global standing, etc. That furthers a cascading effect that’s very negative.
This process is already well underway. The current massive over investment into the AI hype bubble is the final death throes of a failed economy trying to keep its head above water.
A KGB spy and a CIA agent meet up in a bar for a friendly drink
"I have to admit, I'm always so impressed by Soviet propaganda. You really know how to get people worked up," the CIA agent says.
"Thank you," the KGB says. "We do our best but truly, it's nothing compared to American propaganda. Your people believe everything your state media tells them."
The CIA agent drops his drink in shock and disgust. "Thank you friend, but you must be confused... There's no propaganda in America."
so what? this has nothing to do with them taking from those that took before them. china censoring information they do not like is nothing new and seems irrelevant to this conversation
Fair enough you don't care, some people might want to know about the Uyghur happy camps, mass organ harvesting and such.
In a world where information sources are only going to dwindle, it is not in anyone's interest to empower actors that will use these to manipulate perceptions
The same people who say there is a Uyghur genocide are the ones who deny there is a Palestinian genocide. Guess which one there's video evidence of (a horrific, endless amount of evidence).
And some people might want to know about [insert your favourite beyond-the-pale-in-the-US topic here, we're on a US forum after all]. I think it would be great if those people could turn to Chinese models, while anyone wanting to know about Uyghur camps can ask the US ones.
There aren’t any. The idea that the censorship environment is the same in the US as it is in China is nonsense used to ‘both sides’ away concerns about the CCP.
The models to themselves are currently generally not censored though and the guardrails are on the application layer?
Of course that might change in the future but as long as the Chinese companies continue publishing their research and models it only makes it easier for third parties to catch up with them.
Can you name one? It's an honest question, first of I'm not American, but also the stuff I do come up with (asking an LLM how to blow up a school or whatever) would also be handled similarly in China, so those would be a wash, and I can't think of any that aren't.
I would imagine a lot of things touching upon progressive politics would be affected (there were a few high-profile incidents demonstrating bias like Google's black Wehrmacht soldier pictures, but has anyone rigorously tabulated how the various commercial LLMs respond to questions about the gender binary or heritability of human traits considered good or bad?). Overall, I'm too reluctant to even write out in the abstract sequences of words that I never want to explain to a future job interviewer or HR employee who used GPT-7 to comb the internet for all text that stylistically can be traced to me, but just imagine whatever you believe to be vile and wrongheaded opinions in that general space which it is certainly more than justifiable to prohibit. The things that make you think "banning this is good actually" are exactly the things most likely to be banned (and this is true in China too).
Another thing I would try if I had access to the models and enough proxies to hide behind is asking for advice on software/movie piracy or seeing to what extent the models can be elicited to straight up argue against the validity of intellectual property, though there it seems more probable to me that the US models would be permissive.
What historical fact is the US trying to systematically suppress, is the question. This beating around the bush about things that are "considered uncouth in some circles" just underlines there either aren't any, or they're so perfect at it nobody knows of any. So let's just admit than and move on, because what you just said is true about any society, ever.
> The things that make you think "banning this is good actually" are exactly the things most likely to be banned (and this is true in China too).
Then make the case why it is better for the world that the Tiananmen Square massacre is memoryholed. You said A, now say B.
In the meantime, I'll make the case against it, and it's simple: totalitarian control of historical truth, by definition, to be 100% watertight, needs control of the whole globe. It doesn't mean everything needs to be controlled, it means everything needs to be controlled by at least an entity that cooperates on this matter. I.e. another totalitarian bloc.
That makes the CCP, just by their insistence about Tiananmen -- nothing additional required, at all, they could not harm a fly and have no prisons and it would make zero difference -- an enemy, a threat marching towards any thinking human who wants to have agency and dignity. Actually, it's more like a river flowing to the ocean, people in the CCP can have lofty ideals about honesty and factual truth, the system they require to survive in turn requires this to survive, as it is. It requires human spontaneity and human freedom to be dead, completely. That is what totalitarianism is.
And by the same token, we must be wary of those who take that lightly. A friend falling asleep at the wheel will kill you just the same as an assassin who messed with your car.
That is not "as opposed to the US or the EU or Russia or North Korea". It is strictly in addition. No other regime can behind another regime or use it as an excuse. Using the US to distract from the CCP is as odious as doing the reverse.
> I never want to explain to a future job interviewer or HR employee who used GPT-7 to comb the internet for all text that stylistically can be traced to me
So you want to work in an office wearing a tie. Well, people who want to work as roadies and fight all night would never hang with you if they found out. And if that was your priority, you would be more afraid of NOT being blacklisted by the corpos than not being part of them.
But seriously, are you saying you are as afraid of being found out as the person you are, because you're not already wearing that on your sleeve, in exactly the same way as people who have to fear that they and their family get disappeared, tortured, for remembering friends that got murdered? That would be totally a you problem.
That's the rub. Today on the internet, no one can validate that there exists any "trustworthy search". If we don't control the servers and validate the software on it, then we are taking someone's word for it. Most often that someone is a marketing person at the hosting organization. Occasionally it's a random comment on the internet behind a pseudonym we know nothing about.
People are becoming more and more, justifiably, cynical. And it's good for society that they are.
> If we don't control the servers and validate the software on it, then we are taking someone's word for it
That's what trust is, though. I find Kagi worthy of trust since I have no evidence that they are doing anything that would violate my trust in them, and their incentives align with mine. I give them money, they give me good search results.
> People are becoming more and more, justifiably, cynical. And it's good for society that they are.
I do not believe that this is good. Democracies are built on trust. Without trust, they die, as we can plainly see in the US right now, where people are so distrustful of each other that it is starting to destroy the very fabric holding together society itself.
The entire point of the Constitution was to not trust each other. We wouldn't even need checks or balances if trust were a reasonable governing philosophy. Nor a Bill of Rights, nor any explicit admonition against slavery.
The reason we need these and many more is that we don't trust each other. And it's wise that we don't.
This doesn't mean trust is bad, only that the founders, rightfully, knew that it leads to complacency.
What's landed us where we are is not a lack of trust. (We've lacked trust from day one, that's how we got the Constitution in the first place). What's landed us here is complacency.
It's distrust - in foreigners, in each other - that got extremists elected, and it's distrust - in the media, in the institutions - that gives them unchecked power to destroy the country.
None of which would have happened if we had not gotten so complacent about defending our democracy through vigorous enforcement of the Constitution, and voting in a fashion more informed by wisdom than emotion.
At the root of all you lament is not lack of trust, it is complacency. Decades long complacency.
Lack of trust is what our republic was founded on, and the entire reason it was structured the way it was. Our republic is specifically designed to function in a trust-free environment.
But no republic, nor any other organizational construct, no matter how well designed, will survive complacency.
Kagi's business model also aligns incentives better than an ad and/or data mining fueled business model. Of course, this isn't proof, but it is evidence. One can also try the search in areas that one knows deeply and compare it to other search engines to see if it's truly better (I have not done this, but it is possible and isn't just trusting marketing.).
I don't use Kagi primarily because I'm quite averse to subscriptions if I can find another way.
Just Devil’s Advocate, but couldn’t you ask these models about controversial topics in US or Chinese history. Like, what happened at Tiananmen? Or what happened during the Tulsa riots?
Aren't these the same guys who won't even let us have Chinese cars?
I'm not as confident as you that they will keep allowing us access to technology as strategic as AI models out of China and elsewhere that undercut US models in the market.
To everyone reading, download open models from anywhere as soon as they are released. You really have no guarantee at all that access to those models won't be cut off in the future with the stroke of a President's pen. Those downloads are your insurance policy. You'll always be able to access whatever you've already downloaded.
People move towards places with growing economic opportunities.
I hope people aren't expecting data centers to provide "growing economic opportunities". That's not really what data centers are about.
Data centers are infrastructure in the same way nuclear plants or canals are infrastructure. Water infrastructure carries the Colorado to Phoenix and other areas in the West. Unfortunately, this does little for people in Colorado. The idea is that the benefit of feeding water to people throughout the west is worth the cost of building and maintaining extensive water infrastructure.
AI infrastructure should be thought of in the same manner. If you're going to have a requirement that data centers provide all these jobs in the places they're built, then data centers are never going to be able to get out from under the PR hammer. And most citizens are going to continue to be disappointed. Because they're thinking about it wrong.
Not that what you've written isn't the case. However, in addition to what you've written, (or probably even before what you've written), there's the fact that everything they're training on is stolen IP. Same with US LLM labs.
Let's not kid ourselve's about where the training data is coming from. They are not asking artists, writers, coders, content creators, etc etc etc for permission to use their creations.
Anthropic, Moonshot et al are doing incredible things, but we shouldn't gloss over the costs. Both present and future costs are kind of enormous.
reply