So this means that China’s low cost models won’t benefit the rest of the world? Then we have to make our own. If we only have expensive models and they have cheap ones they will be at a huge advantage not unlike low cost of labor advantages.
It's wild how restricting China's access to our GPUs led them to create some of the top & most widely used open source ai models, while the US's models seem to be few and far between.
It looks like Meta has given up on Llama, which was one of the most popular, in favor of going proprietary to make more money.
This also leads me to believe the discussion about China slipping propaganda into their models (the whole "ask it about Tiananmen Square" thing) was never really a thing or at least not a priority for them.
Turns out, prohibition doesn't work in any of its many forms. This has been obvious to anyone with a semblance of critical thought for the last 100 years, but the US government never seems to learn.
> prohibition doesn't work in any of its many forms.
It definitely does work in that it will prevent us customers from accessing cheap goods from China and thus enrich the local AI companies that are paying the politicians to ban the Chinese models.
It doesn’t mean that. Low-cost models will still enter the market because China isn’t the only country capable of creating them. If you want to do the manufacturing comparison then it’s like the latest iPhone versus the cheap Android phone from Wal-Mart.
They are both phones and do all the same things but you buy one and you pay a lot more money for it because at the end of the day it is way better. Cheap models capture the slim margin and marginal utility product mix, whereas these “high-end” models capture more of that iPhone market.
China doesn’t want to just be a low-cost manufacturer (you can see this reflected in the ongoing hollowing out of Germany’s economy losing 10,000 manufacturing jobs/month) - they want to make high end models too because the economics and capabilities are far superior.
This doesn’t hold up imo. Your phone analogy fits the description of a harness. That is the thing you interact with. The LLM is more like the processor in that phone. As long as it gets the work done, I will choose the cheapest/fastest because I can switch it out in a matter of seconds anyways.
You're straying away from the analogy though. If the LLM is more like the processor in the phone then you'd just compare the Apple series chips to the low cost chips in the low cost phone.
> I will choose the cheapest/fastest because I can switch it out in a matter of seconds anyways.
Sure but that's why you would go with the Apple series chip in this analogy. It's more expensive but when it comes to getting what you want done it would actually be cheaper and faster.
For some reason everyone wants to praise China's low-cost models and simultaneously applaud their move to create and restrict higher-cost/performative models while criticizing the US for also creating and restricting higher-cost/performative models. It really doesn't make sense.
If you want to say all models are the same or close enough so you'll just switch to the cheapest and fastest then market forces will allow American companies to just create those same models or, since they're open source, they'll just use those but American companies can just do that too and they're choosing not to because performance matters, approximate winner-take-all matters, and cost is important but not the only factor.
If you want to suggest that America is doing all the high cap-ex spend and then China is just undercutting with cheap models, then American companies will adopt that strategy too if it is what is most profitable and then Chinese companies will have to spend on cap-ex to get ahead and they can be undercut by American companies. But this doesn't reflect reality and China's actions demonstrate that point because they're still spending cap-ex and trying to produce higher-end models which they then restrict and blah blah blah.
Sad to see this. They were the true disruptors to the space and gave the open source and home lab communities the opportunities to do things OpenAI and Anthropic would never allow. It also might have been the biggest factor in deflating the AI bubble we're in.
So this means that China’s low cost models won’t benefit the rest of the world? Then we have to make our own. If we only have expensive models and they have cheap ones they will be at a huge advantage not unlike low cost of labor advantages.
It's wild how restricting China's access to our GPUs led them to create some of the top & most widely used open source ai models, while the US's models seem to be few and far between.
It looks like Meta has given up on Llama, which was one of the most popular, in favor of going proprietary to make more money.
This also leads me to believe the discussion about China slipping propaganda into their models (the whole "ask it about Tiananmen Square" thing) was never really a thing or at least not a priority for them.
Turns out, prohibition doesn't work in any of its many forms. This has been obvious to anyone with a semblance of critical thought for the last 100 years, but the US government never seems to learn.
> prohibition doesn't work in any of its many forms.
It definitely does work in that it will prevent us customers from accessing cheap goods from China and thus enrich the local AI companies that are paying the politicians to ban the Chinese models.
It doesn’t mean that. Low-cost models will still enter the market because China isn’t the only country capable of creating them. If you want to do the manufacturing comparison then it’s like the latest iPhone versus the cheap Android phone from Wal-Mart.
They are both phones and do all the same things but you buy one and you pay a lot more money for it because at the end of the day it is way better. Cheap models capture the slim margin and marginal utility product mix, whereas these “high-end” models capture more of that iPhone market.
China doesn’t want to just be a low-cost manufacturer (you can see this reflected in the ongoing hollowing out of Germany’s economy losing 10,000 manufacturing jobs/month) - they want to make high end models too because the economics and capabilities are far superior.
This doesn’t hold up imo. Your phone analogy fits the description of a harness. That is the thing you interact with. The LLM is more like the processor in that phone. As long as it gets the work done, I will choose the cheapest/fastest because I can switch it out in a matter of seconds anyways.
You're straying away from the analogy though. If the LLM is more like the processor in the phone then you'd just compare the Apple series chips to the low cost chips in the low cost phone.
> I will choose the cheapest/fastest because I can switch it out in a matter of seconds anyways.
Sure but that's why you would go with the Apple series chip in this analogy. It's more expensive but when it comes to getting what you want done it would actually be cheaper and faster.
For some reason everyone wants to praise China's low-cost models and simultaneously applaud their move to create and restrict higher-cost/performative models while criticizing the US for also creating and restricting higher-cost/performative models. It really doesn't make sense.
If you want to say all models are the same or close enough so you'll just switch to the cheapest and fastest then market forces will allow American companies to just create those same models or, since they're open source, they'll just use those but American companies can just do that too and they're choosing not to because performance matters, approximate winner-take-all matters, and cost is important but not the only factor.
If you want to suggest that America is doing all the high cap-ex spend and then China is just undercutting with cheap models, then American companies will adopt that strategy too if it is what is most profitable and then Chinese companies will have to spend on cap-ex to get ahead and they can be undercut by American companies. But this doesn't reflect reality and China's actions demonstrate that point because they're still spending cap-ex and trying to produce higher-end models which they then restrict and blah blah blah.
Sad to see this. They were the true disruptors to the space and gave the open source and home lab communities the opportunities to do things OpenAI and Anthropic would never allow. It also might have been the biggest factor in deflating the AI bubble we're in.
As long predicted.
In 1-2 years all AI labs will be nationalized, put under military control, and models will be guarded like nuclear weapons.
"You were the chosen one! You were supposed to destroy them, not join them!"