the sovereign AI thing is fascinating to me for many reasons:
* it's time to build out stacks! servers, deployments, things! Especially in Europe. Hence sovereign AI is already a big sales buzzword for lots of hardware and hardware adjacent companies - NVIDIA (obviously), Cisco, Oracle and my own employer, Red Hat (hooray!)
* It makes the whole "AI is a bubble" thing completely irrelevant. OpenAI and Anthropic are no longer relevant, AI models themselves no longer need to "make money", and hopefully lots of really good / free ones start becoming commonplace. Ed Zitron goes out of business.
* How fascinating that Trump/ Hegseth maybe didnt completely create this but boy did they throw a thousand gallons of gas on the fire, bullying Anthropic around and scaring the crap out of every government, institution and business that's not 100% US only who is using Claude. nice job idiots!
* Can models be maybe trained with little more focus on only what they need to know? They need to know code, math, business, science. They don't need to know how to rip off movies or art. These LLMs would hopefully not be spamming Amazon with shitty books and Instagram with fake influencers (but I'm likely being too hopeful)
m8ven•1h ago
Many good points, but this may also result in a proliferation of more "funded/established" AI choices or many more open source choices, making this field more competitive and less efficient. And that could drives up the bubble more
zzzeek•1h ago
* it's time to build out stacks! servers, deployments, things! Especially in Europe. Hence sovereign AI is already a big sales buzzword for lots of hardware and hardware adjacent companies - NVIDIA (obviously), Cisco, Oracle and my own employer, Red Hat (hooray!)
* It makes the whole "AI is a bubble" thing completely irrelevant. OpenAI and Anthropic are no longer relevant, AI models themselves no longer need to "make money", and hopefully lots of really good / free ones start becoming commonplace. Ed Zitron goes out of business.
* How fascinating that Trump/ Hegseth maybe didnt completely create this but boy did they throw a thousand gallons of gas on the fire, bullying Anthropic around and scaring the crap out of every government, institution and business that's not 100% US only who is using Claude. nice job idiots!
* Can models be maybe trained with little more focus on only what they need to know? They need to know code, math, business, science. They don't need to know how to rip off movies or art. These LLMs would hopefully not be spamming Amazon with shitty books and Instagram with fake influencers (but I'm likely being too hopeful)
m8ven•1h ago