Like you said, even with DeepSeek's impressive efficiency gains, they're still constrained by computing architecture. Though I wonder if this might push innovation in unexpected directions, kind of like how resource constraints in mobile development led to entirely new approaches.
I'm curious how you see the democratic/autocratic AI development split playing out. Some folks in my network argue it's the defining factor, while others see it as less significant than raw technical capabilities. Only time will tell John.
Thanks for your thoughts, Neela! I’m more of an outsider to the space (but very curious observer) so it’s interesting hearing your take.
As someone more from the political and legal world, however, I see the democratic/autocratic divide as the defining factor. I’m not sure how large language models or AI systems in general can progress if they’re constantly constrained by government putting its thumb on the scale of innovation and speech.
China only really started to grow once it opened up its economy and became quasi-capitalistic in the late 20th century. The history of innovation largely stems from more open societies that value the rule of law, independent judiciaries, and free speech.
A big factor in AI innovation will be its government use cases. A government that uses it for surveillance and limiting civil liberties compared to one that promotes innovation while ensuring a baseline of safety and security is going to significantly deter the bold entrepreneurism that’s needed to advance AI to the next level.
Founders prize certainty, both from a legal and economic perspective, more than almost anything. If government is constantly looking over the shoulder of a founder or company, there’s limited freedom to innovate.
So while I’m very impressed with what DeepSeek has accomplished, I’m deeply skeptical that they’ll be able to maintain any advantage long term given autocratic government constraints. That’s assuming, of course, that the US and the West in general doesn’t lose sight of the rule of law and let’s innovators innovate instead of overburdening them with regulations (although I do think uniform minimum standards are needed).
This took me back to how Bell Labs vs Soviet research programs played out during the Cold War. Same brilliant minds, and wildly different results. I'm seeing companies having to essentially build dual architectures - one for domestic compliance and another for international markets. The computing and data requirements make this exponentially more complicated than traditional software development.
What's interesting about DeepSeek's approach is how their resource constraints might actually drive innovation in unexpected ways. Some of our clients are already exploring similar efficiency techniques, though without the same political restrictions. Because compliance is an issue in the West though I feel like with the new administration - the desire to win may be greater. I will write an article soon - thank you John.
This has been the buzz in my circle today.
Like you said, even with DeepSeek's impressive efficiency gains, they're still constrained by computing architecture. Though I wonder if this might push innovation in unexpected directions, kind of like how resource constraints in mobile development led to entirely new approaches.
I'm curious how you see the democratic/autocratic AI development split playing out. Some folks in my network argue it's the defining factor, while others see it as less significant than raw technical capabilities. Only time will tell John.
Thanks for your thoughts, Neela! I’m more of an outsider to the space (but very curious observer) so it’s interesting hearing your take.
As someone more from the political and legal world, however, I see the democratic/autocratic divide as the defining factor. I’m not sure how large language models or AI systems in general can progress if they’re constantly constrained by government putting its thumb on the scale of innovation and speech.
China only really started to grow once it opened up its economy and became quasi-capitalistic in the late 20th century. The history of innovation largely stems from more open societies that value the rule of law, independent judiciaries, and free speech.
A big factor in AI innovation will be its government use cases. A government that uses it for surveillance and limiting civil liberties compared to one that promotes innovation while ensuring a baseline of safety and security is going to significantly deter the bold entrepreneurism that’s needed to advance AI to the next level.
Founders prize certainty, both from a legal and economic perspective, more than almost anything. If government is constantly looking over the shoulder of a founder or company, there’s limited freedom to innovate.
So while I’m very impressed with what DeepSeek has accomplished, I’m deeply skeptical that they’ll be able to maintain any advantage long term given autocratic government constraints. That’s assuming, of course, that the US and the West in general doesn’t lose sight of the rule of law and let’s innovators innovate instead of overburdening them with regulations (although I do think uniform minimum standards are needed).
It will be a very interesting space to watch!
This took me back to how Bell Labs vs Soviet research programs played out during the Cold War. Same brilliant minds, and wildly different results. I'm seeing companies having to essentially build dual architectures - one for domestic compliance and another for international markets. The computing and data requirements make this exponentially more complicated than traditional software development.
What's interesting about DeepSeek's approach is how their resource constraints might actually drive innovation in unexpected ways. Some of our clients are already exploring similar efficiency techniques, though without the same political restrictions. Because compliance is an issue in the West though I feel like with the new administration - the desire to win may be greater. I will write an article soon - thank you John.