We do know that it was massively oversold. Elon claimed that he could essentially balance the budget off of DOGE's work by eliminating "fraud, waste, and abuse." Well anyone remotely numerate knew that was impossible, and it didn't happen. Not even in the same ZIP code, numerically speaking.
The line of business is fine. Tesla & SpaceX both make things people want. Neither cars nor rockets are being disrupted by AI. They could achieve good-to-great growth under stable, full-time, drug-free management. But if you feel you need a large amount of capital and you want to retain control of SpaceX and Tesla, bolting these startups onto them is a way to double down. It's just a shame to see those existing things get torched.
I ran into (2) when trying out nanoclaw many months ago (turned out to be useless, I ditched it). I wanted an LLM to sort my email like an EA might, what I got was an LLM writing regexes to attempt to sort my email. If LLMs have been trained to be the ultimate programmers, the one pitfall is that the best programmers tend to be lazy.
People always find new ways to make mistakes. The approach in aviation is to investigate, identify root causes, and do what is possible to prevent the same mistake happening again.
There is no guarantee we'll ever reach perfection.
It's more about managing mistakes. Pilots and other people in aviation still make a lot of mistakes all the time. But aviation figured out how to deal with that fact. Design procedures in a way, that mistakes self-correct most of the time. Add sanity checks exactly where they matter, and many other strategies.
I remember when SO had codegolf on the main site. Maybe if they weren't such killjoys people would have still been interested in participating in some programming oriented community even though AI displaced the actual writing of computer languages.
Sadly the bullshit asymmetry principle has been supercharged in the AI era though I think many imagined it would be the AI's doing rather than jealous, paranoid humans. Now we have incorrect but widely applied folk wisdom that OAI's tech is just rehashing old research.
Which is fine for a video game. Not at all fine if you want to train robots with it. Trying to combine both use cases into a single model is doomed to failure IMO.
I would not be so opinionated, modern coding agents derive from chatbot technology. Very different problem domain. They share language but language is universal so that doesn't say much. Then they figured out that the pretraining is a good starting point for RLVR. Who's to say a more cinematic imagination can't be a good starting point for a world model? It doesn't need to be the whole solution.
There's a large logical leap from "we have very useful raw research data" to "we should give this data only to a small group of people." They explain that "the collaborations empower experts to foster unpacking the results and contextualizing them in the broader academic community" but we are left to wonder why the experts need to be empowered to foster anything. In the case of immediate release the experts will have their hands on the same data as the rest of us peasants, and then they can go right on fostering.
