Airbnb's Chesky Says US 'Misunderstanding' Use of Chinese Open-Source AI Models
bloomberg.com
bloomberg.com
AI is energy intensive so being the lowest cost provider will likely hinge on energy costs.
Non-renewable, carbon based energy is more expensive and thus a competitive liability in the AI race.
In other words, China gets it, the USA doesn't.
https://www.energyprices.net/renewables/chinas-clean-energy-...
Interesting
You can see the effect in the air pollution monitoring, China has reversed course on this front.
As it pertains to the conversation, if you are allowing China to make advances and change you also have to allow that for America - in other words we can compete and build out our grid too and compete across the spectrum, low-cost, energy efficient models, or otherwise.
On the high end China does not currently have a competitor so America stands to compete on both fronts with a current advantage on the high end which is what matters most.
If China commoditizes their models they, well, become commodities which means lots of competition and market entry by competing firms. American firms for example can also create low-cost, low-energy models it’s just that there is no economic model to do so. Racing to the bottom is a fool’s errand.
"Just"? Where do I start? That just outweighs all other problems combined, by far.
So sure you can churn out free/cheap stuff, but that's not a business model, that's desperation. This whole thread and conversation is exactly proving the point.
US - gargantuan amount of spend and expansion of compute power and energy build out with the aim to bring about nothing less than the Machine God. Companies are paying fortunes to run the latest and greatest models and seeing productivity and efficiency gains and increases in revenue.
China - lots of idle clean energy production potential due to planned economy edicts (where have we seen this before, cough housing cough), reliant on American tech stacks and chips, releases free and open source models for folks to tinker with that are behind and continuously behind American models. Oh and nobody is paying for the Chinese stuff.
Here's the real problem: in America kids are booing AI models and demonstrating against tech companies while in China everyone is all-in on AI and fearful of being behind. We're completely and demonstrably capable of dominating this space. Newsflash - America can and does make great stuff. We are incredibly competitive, dynamic, innovative and operate with very deep capital markets. We have world-leading companies and will continue to do so even as China and companies from other countries continue to compete hard too.
The risk factor is anti-American propaganda and leftist subversion convincing people that adoption of technology is a bad thing and then we start banning technology or implementing other ass-backwards policies and we wind up behind.
I do not believe this to be true in the long run. I think American Ai companies understand this. I've already made the switch because the difference is increasingly small. Progress up to now shows us that the small models are about 1 year behind frontier, and at some point will become good enough for the majority of tasks, i.e. you don't always need the biggest or the best. This conversation is already happening, where people are choosing the smaller models from American companies, or moving to open weight models they can run themselves. Heck, Congress even has American companies coming in to testify about their usage of Chinese models because they do not understand the difference between downloading and running it yourself vs making API calls to servers in another country.
I think there's plenty of room for lots of approaches. But similar to how Open Office is nice and all, Microsoft Office makes however many hundreds of millions. I'm not really sure why folks believe that what we're seeing today will be any different. Bleeding edge compounds gains and advantage. American tech companies are run by some of the smartest people on the planet and from all over the planet. They're not spending hundreds of billions of dollars because they're wrong and because some 1-year behind model is a big threat.
There is a likely world (imo) where Big AI produces mega models primarily to distill them down into economically viable sized models
China also has mega models, US also has open models. Both groups are doing the same things. AllenAi is the most open but their models have not been competitive
I personally prefer the smaller models because they are much faster for almost the same quality. I prefer open weights even more for the freedom to use as much as my hardware will let me (~250M tok/day right now) and the expectations they will continue to improve over the coming years. We are going to be running experiments at work to determine what / how much it would take for self hosting a deep seek / kimi to have a nice ROI. A few of us now have the spark to get a taste and catch the bug
Framing it as "desperation" does have some dramatic effect but it says nothing about the potency of the counter-measures being deployed.
> We're completely and demonstrably capable of dominating this space.
That's the kind of blind faith mixed with self-deception that led to the recent war. Depending on who you ask, "we" are completely dominating "them" but the vast majority of "US" aren't dominating anything, we are being dominated by high gas prices and inflation.
Kinky dominatrix games are full of risk and they aren't worth even 10% of their cost.
> We are incredibly competitive, dynamic, innovative and operate with very deep capital markets.
If you knew something about economics, you'd know that the presence of high inflation is a sign of depleted capital markets. Blind faith and reality aren't the same.
> The risk factor is anti-American propaganda and leftist subversion convincing people that adoption of technology is a bad thing... or implementing other ass-backwards policies and we wind up behind.
That's the usual war propaganda, full of meaningless generalities, sound bites and straw men. The value of technology doesn't exist outside of economics - it's the difference between what it provides and what it costs over the time frame of adoption.
In other words, rushing a generally good technology can turn it into a monster - that's exactly what's happening here. It's similar to rushing into war while other means are neglected or deliberately ruined. In both cases the only results are market manipulation, enrichment of the very few and making life harder for the rest.
I don't think it's any less potent than other claims that have been made. When you're behind you tend to release stuff for free (Meta comes to mind).
> That's the kind of blind faith mixed with self-deception that led to the recent war.
I'm not sure why you're bringing Iran into this because I'm not too sure it's relevant, but even so the Iran war was a good idea. You need to free yourself from the short-term political oriented thinking and instead think about this more long-term, strategic, and historical. Iran is firmly in the camp of China and Russia. What better time than now to knock them down severely so that if war comes with one or both China and Russia we have a less capable Iran causing chaos in the Middle East and further disrupting global oil supplies?
> Depending on who you ask, "we" are completely dominating "them" but the vast majority of "US" aren't dominating anything,
I don't know what this means but I guess it sounds cool. Broadly speaking I did say we were capable of dominating in this space, it's not a sure thing which I would be among the first to admit.
> we are being dominated by high gas prices and inflation.
There's a lot to unpack here, but suffice to say it's worse for everyone else, including China. MAGA coal rollers pay more at the pump. Whatever. No big deal.
> If you knew something about economics, you'd know that the presence of high inflation is a sign of depleted capital markets. Blind faith and reality aren't the same.
Can you elaborate? I don't know anything about economics as you said so I'm struggling to understand how "high" inflation of a few percent means the US has depleted capital markets. When were they depleted? For how long will they be depleted? Are you able to answer these questions?
> That's the usual war propaganda, full of meaningless generalities, sound bites and straw men.
Well it's not really propaganda. There was a thread about it on the front page the other day about college students booing commencement speakers when they talk about artificial intelligence.
> The value of technology doesn't exist outside of economics - it's the difference between what it provides and what it costs over the time frame of adoption.
Can you be more elaborate? I'm not sure what you're talking about. I don't know much about economics.
> In other words, rushing a generally good technology can turn it into a monster
Sure! So make sure you are continuously clamoring from every corner that China must slow down its adoption of AI.
I don't see any signs that China is rushing AI adoption, they are simply trying to stay close enough to the US to counter the obvious and loudly proclaimed existential threat of external domination, words supported by wide-ranging sanctions and tariffs.
> I'm not sure what you're talking about. I don't know much about economics.
That's not unusual and I can't blame anybody, except the professional economists, for the sordid state of econ education in this country. Unfortunately, that doesn't make it less of a deal breaker, one cannot claim to understand the political landscape without understanding economics.
Nevertheless, I'll give you a few hints - the rushed buildout of data centers stuffed with expensive GPUs, made in Korea and Taiwan, strains both the semiconductor and energy markets, the prices go up but new production is slow to come and always insufficient. Little or even nothing is gained from the rush, except inflated asset prices (market manipulation) and general inflation.
At the same time, the rest of the economy is starved for capital which goes to the semiconductor makers in Korea and Taiwan to pay for $ 1 million bonuses, per worker, per year. The picture is similar on the market for power equipment. War and Hormuz are the icing on the cake.
In other words, the rush causes a spike in demand which causes a spike of inflation, like about 500% for RAM and somewhat less for GPUs, the high costs require more capital for the same assets, which depletes the capital markets and the continuous expansion of demand with stagnating production leads to an inflationary spiral.
The rushed adoption also causes friction on the labor market where subsided AI is used to fire large number of workers. Looks good in the earnings reports (market manipulation again) but doesn't do anything good for the US economy or population. In this situation, blaming the students cannot be accepted seriously.
If the US is rushing and China is trying to stay close then it makes sense for them to slow down. You can't have your cake and eat it too. Separately if you're claiming that there's an existential threat of being dominated by the US going too fast you admit both that the US is making significant and important progress on AI which, as the trailing nation, would be concerning, and secondly you admit that the US must also push far and fast because to fall behind simply puts it in China's current position which... is trying to catch up for precisely the same reasons.
Your rationale/criticism here doesn't add up.
Regarding tariffs, I don't want to re-litigate, though I will if you want, the thousands of conversations that have taken place, but suffice to say China has not been a good actor and has failed to open their markets. Not only the United States but the European Union have taken action specifically against China because of anti-competitive policies.
We could quibble about the effectiveness of tariffs - seemingly very effective with respect to China, but it would be unfair to suggest that they sprang up out of nowhere and weren't a response to other actions.
> At the same time, the rest of the economy is starved for capital which goes to the semiconductor makers in Korea, Taiwan to pay for $ 1 million bonuses, per worker, per year. The picture is similar on the market for power equipment. War and Hormuz are the icing on the cake.
That just seems like a rather routine labor dispute. Workers want higher shares of the profits. So what?
> The rushed adoption also causes friction on the labor market where subsided AI is used to fire large number of workers. Looks good in the earnings reports (market manipulation again) but doesn't do anything good for the US economy or population.
I'm still not seeing any evidence that the US capital markets are dried up. What source are you pulling from? How much are they down by?
From what I can tell from example reports like this: https://www.pwc.com/us/en/services/consulting/deals/us-capit... things are looking pretty good.
> Looks good in the earnings reports (market manipulation again)
Which specific earnings reports did you read and are citing and claiming to be manipulated? I'd like to read those too.
> In this situation, blaming the students cannot be accepted seriously.
I'm not blaming the students, I'm saying the approach is wrong. If they (and others) fail to adopt AI then the US falls behind China. Or is being behind not a problem? If it's not a problem, why then would it be a problem for China?
Do you understand the difference between being a target of domination and falling behind? Replacing the words of others with your own straw men is a sign of a mind poisoned by ideology.
China has always been behind the US, they've provided the US population with a lot of affordable goods and made some US corporations very rich. Guess what, they've never had a problem with it until they became a target of bullying and talk of domination. And even then, they are quite reasonable and even-handed.
> That just seems like a rather routine labor dispute.
The thought of looking at this as a "routine labor dispute" never even crossed my mind. I'm talking about efficiency of capital, that "dispute" is how you learn about the wasteful use of US capital, information that you can only get by reading between the lines.
The US would be better off and further ahead if China was a market participant and we didn't have to pay exorbitant prices for hardware. In other words, if we haven't started a trade war with China, but rather used them to gradually build the necessary generation and semiconductor capacity, we would be further ahead of them, and they wouldn't mind it.
Please cite the specific words that I replaced.
> The thought of looking at this as a "routine labor dispute" never even crossed my mind.
That's strange, it should have been the first thing that crossed your mind with your extensive economics education. Some Samsung workers wanting more of the share of a company's profits isn't any different than some Teamsters strike or UAW.
> I'm talking about efficiency of capital, that "dispute" is how you learn about the wasteful use of US capital, information that you can only get by reading between the lines.
Sorry, I'm not educated in economics. Reading between the lines isn't something I can do so you'll have to spell it out point by point.
> > China has always been behind the US, they've provided the US population with a lot of affordable goods and made some US corporations very rich. Guess what, they've never had a problem with it until they became a target of bullying and talk of domination. And even then, they are quite reasonable and even-handed.
> The US would be better off and further ahead if China was a market participant and we didn't have to pay exorbitant prices for hardware. In other words, if we haven't started a trade war with China, but rather used them to gradually build the necessary generation and semiconductor capacity, we would be further ahead of them, and they wouldn't mind it.
Unfortunately Taiwan stands in the way of all of that. Are you suggesting we should give up Taiwan to repair relations?
While China has indeed made some US corporations wealthy and have provided Americans with cheap plastic goods, it has adverse effects such as draining the US of manufacturing capacity - this is because China maintains a closed market and has come under both American and European tariffs and trade restrictions because of the imbalance. You're being disingenuous, I hope, by pretending to not have an understanding of international and economic relations with China over time and how it has affected American and European economies.
A couple of examples:
China didn't fall behind, they always have been, but are now catching up and likely to pass by.
China did not drain our manufacturing, that was wall street incentives and the corporations. They were the ones deciding.
Do you think America has the capacity to defend Taiwan after what we've seen with Iran? Do you think the people have the will to endure the hardship that comes with that? Our military is currently designed for counter terrorism, not a fight with a near peer. They want to build super planes and battleships, but war has changed.