But they wouldn't identify any specific problems. I.e., they're lying.

The normal strategy is to take them on one at a time...and have the expectation that after you've won a couple, the rest will start to fall in line without being sued.

That's the wrong problem. He's almost certainly planning to loft multiple satellites with each launch. But there are other problems it's reasonable to be dubious about.

I'm going to wait until he's launched a few to form a definite opinion...though I'll admit that I consider it unlikely.

It's possible that he's got an answer to that problem. I'll wait to form an opinion. (Yeah, I'm skeptical too, but I don't think one should be certain based on the available evidence. Also, there are several other potential problems, but maybe he's got answers.)

I think the correct answer is "you don't use a screwdriver as a hammer". Used correctly, AI tools can be quite helpful. But reports seem to show that only around 20% of companies use them correctly. (According to at least one report they produce a 5% improvement for one particular task. Whether that includes the cost of use, the article didn't say.)

You misunderstood me. It was Sony I've been avoiding.

OTOH, Chinese might be better than Sony. They sold their souls so long ago that I can't remember the last of their products I bought. Once upon a time they had a name for quality, but then they sold their soul to Hollywood.

FWIW, primitive savages are generally extremely charitable to others in their tribe. It's people who live in larger groups not as closely related that tend to be more selfish.

Outdated doesn't mean "stops working" or "isn't useful". I'm highly skeptical because of heat problems, and comments others have made about GPUs/TPUs not being space-rated. They sure aren't radiation hardened. But if they keep working, they don't become less useful, just less competitive against the latest model.

What orbits is he asking for? If it's LEO, then they'll burn out rapidly. If it's geostationary, then it's already crowded. If it's everything in between, then there's (potentially) a lot of space, but it's a bit more expensive to get there. And you'd want everything to have a VERY low eccentricity orbit, so then didn't interact.

One good idea would be to physically link the satellites together, but that would require a lot of development that hasn't really been tested. (It would also mean they keep position as long as any of them keep working.)

And how is he planning to address cooling? That's usually quite difficult and expensive on satellites.

Maybe he (or anyway his company) has answers to those problems, but I sure don't.

That NOT a problem. Don't expect any tool to be the be-all-end-all. It's really useful to have a bunch of classes of error be automatically detectable.

You're almost certainly correct. So do large corporations. So do spammers. So, AFAIKT, does the US govt. The constraints they operate under are a bit different, but the attitude seems to be the same.

A quick search didn't reveal what law he broke. Or anything about a trial.

It's too early to be sure how much is going to happen, but it's clear that a LOT is going to happen. The current state is intensely unstable. I *suspect* that AGI will be coming before 2035, but perhaps not. Whether or not it does so, it doesn't change the instability of the current state. Desk jobs that continue to exist are going to change drastically, and we don't yet know in what way. Perhaps AIs will continue to have a really short memory, but perhaps not. Just that one fork in the road will alter things immensely.

You can't have the specific expertise, because he's training it himself. Expect it to catch the kinds of errors he trains it to catch, and to probably do a good job at that...and a worse than lousy job on other kinds of errors.