DennisP

15 ms·

391.
by DennisP 1y ago
Tesla spent about $10 billion on AI training hardware in 2024.[1] They just might not be getting as much bang for the buck, due to their paucity of sensors. Huawei get similar performance without Tesla's four billion miles of training
392.
by DennisP 1y ago
You might not have the classical guardrails, but you are providing the neural net with a lot more information. Even humans are starting to find it useful to get inputs from other sensor types in their cars. I agree that Tesla may have made
393.
by DennisP 1y ago
Huawei has a self-driving system that uses three lidars, which cost $250 each (plus vision, radar, and ultrasound). It appears to work about as well as FSD. Here's the Out of Spec guys riding around on it in China for an hour: https:&
394.
by DennisP 1y ago
ZK rollups which greatly compress data on chain without losing security guarantees, combined with temporary storage using data sampling so each node only has to store a small portion of it. The zk rollups are live and support significantly
395.
by DennisP 1y ago
Also the capacity is significantly higher with L2 included, and increasing rapidly. With zkrollups and a decentralized sequencer, you basically pay no penalty vs. putting transactions on L1. So far I think the sequencers are centralized for
396.
by DennisP 1y ago
That is Stripe's fee. If you use the blockchain directly instead of through Stripe, you don't pay that.
397.
by DennisP 1y ago
Which L1 do you mean? I don't see any fee amounts on Tempo's page. Most stablecoin transactions are on Ethereum and the fees are neither percentages nor fixed dollar amounts. They just have congestion pricing, so it depends on how
398.
by DennisP 1y ago
Businesses do lots of transactions without trusting anyone else's records. Then they do lots of slow, expensive mutual auditing.
399.
by DennisP 1y ago
There's not that much CPU involved. Most of the stablecoins are on Ethereum, and I think the rest are on other proof-of-stake platforms, not Bitcoin.
400.
by DennisP 1y ago
Hopefully for a better reason than the report against John Carmack.
401.
by DennisP 1y ago
Maybe we should think of current AIs as not so much artificial intelligence, as collective intelligence. Which itself can be extremely valuable.
402.
by DennisP 1y ago
Indirectly holding through multiple entities still counts? How would someone legally get around that? It seems like whatever you do, it's still summing up.
403.
by DennisP 1y ago
Using LLMs to build the libraries you need seems like a fantastic way to work with them, since they've probably been trained on code that does similar things.
404.
by DennisP 1y ago
In fact that's Aubrey de Grey's approach: rather than trying to figure out all the complicated processes involved in causing the damage in the first place, so you can slow them down, just directly fix the damage afterwards. There&
405.
by DennisP 1y ago
Generally you don't have to lose your insurance just because you modify your car. You just have to let them know you're doing it, and probably pay a little more.
406.
by DennisP 1y ago
What I mostly see from EVs is varying the number of motors. You get the manufacturing cost reduction on your motors without having to actually over-spec your base model. Even this isn't the whole story though, at least at the high end.
407.
by DennisP 1y ago
Except training data is not the reason. Or at least, not the only reason.
408.
by DennisP 1y ago
I'm not saying it's enough for the outputs to be "equivalent enough." I am saying that if the outputs and inputs are equivalent, then that's enough to call it the same thing. It might be different internally, but
409.
by DennisP 1y ago
Maybe pure language models aren't world models, but Genie 3 for example seems to be a pretty good world model: https://deepmind.google/discover/blog/genie-3-a-new-frontier... We also have multimodal AIs that
410.
by DennisP 1y ago
If a human does A, that required reasoning. Same for AI. If a human does B, that didn't require reasoning. Same for AI. Believe it or not, people do make an effort to test their AIs on problems that they could not have seen in their tr
411.
by DennisP 1y ago
Taking a pragmatic approach, I would say that if the AI accomplishes something that, for humans, requires reasoning, then we should say that the AI is reasoning. That way we can have rational discussions about what the AI can actually do, w
412.
by DennisP 1y ago
Though we should also bear in mind that the wolf in that story eventually showed up.
413.
by DennisP 1y ago
I've lived in Germany where the public transport was fantastic. I used it all the time. It's also a great place for bicycles. And guess what, people still drive cars. There's still plenty of traffic in town and on the autobah
414.
by DennisP 1y ago
But drugged up people in a driverless taxi seems best of all.
415.
by DennisP 1y ago
Maybe the right framing here isn't safety, but how pleasant or unpleasant the experience is.
416.
by DennisP 1y ago
Meanwhile, pretty much everybody else doing safety-critical real-time control prefers multiple sensor types. For an example of that in self-driving, Huawei uses vision, lidar, radar, and ultrasound. Out of Spec let it drive them around for
417.
by DennisP 1y ago
Yes, but we're a long way from putting AI hardware in cars that's as powerful as the human brain. Plus our cameras aren't as good as human eyes.
418.
by DennisP 1y ago
https://www.economist.com/interactive/trump-approval-tracker
419.
by DennisP 1y ago
That's sad, but check recent polls in Trump's approval rating for various issues. He's underwater and trending down, pretty much across the board. https://www.economist.com/interactive/trump-approval-trac
420.
by DennisP 1y ago
Conflicting data is an issue even when you just have lidar, and it's not hard to deal with. The cofounder of Waymo taught one of the first Udacity courses on this subject. He went through a small Python project that processed lidar poi

More ›