Rendered at 00:50:04 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
palata 14 hours ago [-]
The US government has had access to all the data collected by the US BigTech for a long time, and now even more with AI. This is most definitely a digital sovereignty problem.
> You essentially pay for intelligence twice, once with money, and again with something even more valuable: the proprietary knowledge you must reveal to make that intelligence useful.
Coming from a company that has made a business out of collecting the proprietary knowledge their users must reveal to use their services, I think it has value.
> Nadella argues that if AI companies get to freely scrape the internet to train their models, it’s only fair that enterprises get to study — or “distill” — those models in return.
I totally agree with that. My hope is that open models (distilled from proprietary models if needed, I don't mind) will be accessible enough to have real competition on the services. It should be possible for a company to use an AI based in their jurisdiction instead of leaking all their data to the US or China.
nlpnerd 10 hours ago [-]
"Nadella argues that if AI companies get to freely scrape the internet to train their models, it’s only fair that enterprises get to study — or “distill” — those models in return."
Two wrongs don't make one right
palata 8 hours ago [-]
I agree as well... doesn't seem incompatible with what I said, though?
ggm 14 hours ago [-]
The loss of privacy in the data worries me immensely. I would be amazed if there are not significant public/open data analysis opportunities into government strategic planning which in effect reconstruct content never meant to have left a secure facility.
palata 10 hours ago [-]
> data analysis opportunities into government strategic planning
To be fair, governments can already ask companies to give them access to their data. Now given that most companies rely on BigTech for pretty much everything, it means that the US can access to all that and use it for whatever they want (e.g. help one of their big companies win a contract). This is already a strategic risk that has existed for years (decades?).
As a hypothetical example (I don't know what Airbus uses), let's say Airbus uses GMail and Teams for communications. The US can access internal communications of Airbus through those US services (which would not be allowed to say it even if they wanted to) and leak information to Boeing, helping them win the contract.
AI just makes it worse, and again the biggest AI companies are in the US.
ggm 1 hours ago [-]
I actually meant the reverse: Government and related information, highly strategic information is dispersed, but winds up in the model. Inferences from the information are not uploaded but the model is trained in ways which expose the inferences, and thus the secure-room-only information of strategic value leaks into the model, and becomes visible to people sharing access to the model. If the partitions between government and private models are weak, then I see the risk.
Quite a lot of government secrets are bound up in trade. If the data about trade and goals in private sector enterprises forms the backbone of department of state strategic intent, then to some extent, the model will reinforce "what is the actual intent of the state department in pursuit of these apparent goals over the same data"
"oh, I didn't SAY that we were going to totally onshore VLSI before 2030" is one thing. Being able to draw the inference over land deals, taxation changes, state level decisions, planning decisions, related strategic supply chain decisions, pending cases in review, matters not available for discussion as a collected set over time...
> You essentially pay for intelligence twice, once with money, and again with something even more valuable: the proprietary knowledge you must reveal to make that intelligence useful.
Coming from a company that has made a business out of collecting the proprietary knowledge their users must reveal to use their services, I think it has value.
> Nadella argues that if AI companies get to freely scrape the internet to train their models, it’s only fair that enterprises get to study — or “distill” — those models in return.
I totally agree with that. My hope is that open models (distilled from proprietary models if needed, I don't mind) will be accessible enough to have real competition on the services. It should be possible for a company to use an AI based in their jurisdiction instead of leaking all their data to the US or China.
Two wrongs don't make one right
To be fair, governments can already ask companies to give them access to their data. Now given that most companies rely on BigTech for pretty much everything, it means that the US can access to all that and use it for whatever they want (e.g. help one of their big companies win a contract). This is already a strategic risk that has existed for years (decades?).
As a hypothetical example (I don't know what Airbus uses), let's say Airbus uses GMail and Teams for communications. The US can access internal communications of Airbus through those US services (which would not be allowed to say it even if they wanted to) and leak information to Boeing, helping them win the contract.
AI just makes it worse, and again the biggest AI companies are in the US.
Quite a lot of government secrets are bound up in trade. If the data about trade and goals in private sector enterprises forms the backbone of department of state strategic intent, then to some extent, the model will reinforce "what is the actual intent of the state department in pursuit of these apparent goals over the same data"
"oh, I didn't SAY that we were going to totally onshore VLSI before 2030" is one thing. Being able to draw the inference over land deals, taxation changes, state level decisions, planning decisions, related strategic supply chain decisions, pending cases in review, matters not available for discussion as a collected set over time...