He did sell billions of dollars of chips to the shell company his spy girlfriend worked for, violating export controls. So there is that.
1,695 karma · joined January 11, 2024
There are so many other children in the world that deserve consideration before his (grown adult, very rich) children.
Jensen should be apologizing to Americans for compromising NatSec.
They lucked into being the available useful product when the transformer dropped. This was not some megabrain multi year preparation they had been doing.
Jensen did attempt to hide the affair. That's why it was an affair. Even now he does not mention the fact that he sold billions of dollars of chips to his girlfriend's shell company, violating export controls and creating a NatSec risk.
If you're asking about my biases: I am significantly invested in AI (all US labs including Anthropic/OpenAI, and the hardware supply chain) because I believe it will be the most transformative tech in human history. I am also American, and so I'd prefer not to have a nation that blends thousands of human rights activists getting to superintelligence first.
My investments in the frontier have grown so I do not claim to be unbiased.
Another bias - I think HN's user base is on average consistently shortsighted when it comes to this technology:
1/ The HN zeitgeist has been consistently wrong about AI since LLMs were introduced. Old HN threads are hilarious time capsules demonstrating how headstrong+dumb the median HN opinion is. These same people post today as if they never said anything dumb a year ago, in the same overconfident and cynical tone.
2/ The median technologist's takes here are actively destructive to the likelihood of us surviving superintelligence. As technologists we all have a responsibility to ensure this goes well - our companies and societies look to our opinions. If people here introspected a bit more and reflected on the previous times they were wrong, and saw this tech for what it really is, we would all have a better chance of surviving superintelligence.
Credit card fraud and the illegal data routing is a crime, but that's something the affected customers would need to sue for.
There is a difference between "white-knighting" and pointing out that maybe businesses shouldn't use an API that has been proven to illegally route data elsewhere, resulting in literal state secrets being leaked.
There is nothing wrong with using DeepSeek's models locally or on your own cloud, and using another search provider! But using DeepSeek's own APIs is pure idiocy unless you really don't care about your data security/privacy (and that of your users).
> And still continue to give them merchant accounts to continue with that?
The AI companies don't steal the credit cards. They instead fund unaffiliated grey-market transfer stations that steal credit cards. They feign compliance by simply funding another transfer station.
DeepSeek could sue for defamation if they are lying. There's a reason they aren't.
Anyways, they can easily be sued by DeepSeek if they are lying.
> critical thinking is enough to know that Anthropic/OpenAI are responsible data sovereign and others are not
Critical thinking is not relying on a confirmed bad actor because you think the others maybe could be bad actors.
I don't think that X post from someone that sells relays is meaningful. Obviously Anthropic is aware of what relays are, they aren't dumb enough to conflate relays with DeepSeek.
"Anthropic/OpenAI bad!" cynicism is not a replacement for critical thinking (as it increasingly seems to be on HN).
You don't buy a hedge fund because it outperforms the S&P, that's just a bonus. You buy one because it has better returns for a given level of risk, and because the returns are uncorrelated with the S&P.
We won't know the answer to this until IPO and until each company reveals the economics of their models (training and serving).
Yes, AI can explore thousands of ideas at once, no, that's not "brute forcing the search space" because the search space is way larger than you think it is.