Companies say all sorts of things which investors don't take at face value, often for the sake of marketing and recruiting. Meta circa 2020 wanted us to believe we'd be swapping NFTs in a virtual mall and having all our meetings strapped into VR headsets by now. That doesn't mean their market cap at the time was based on an assumption they were right, in fact it is much higher now that the metaverse idea flopped. Unlike that situation where money was being spent on projects that turned out to be mostly pointless, AI labs don't have much of a financial stake in a scenario such as "lots of demand for AI, leading to mass unemployment and/or doomsday" as opposed to one like "lots of demand for AI, leading to the world being mostly the same as before but with more AI and expensive RAM".
AI obviously has military applications, no doom required to think the military will want a lot of it. I'm asking about the extreme scenarios like "AI is gonna take everyone’s jobs" and the other "totalizing narratives" that geohot is referencing.
What's the evidence that the doom narrative is connected to valuations? It seems more like a marketing/recruiting strategy. As the post points out, institutional investors generally think the idea of mass unemployment is BS, and they are investing accordingly.
>Framing it as an argument against American manufacturing or jobs is complete nonsense.
It is a strong argument against those things in at least three ways: (1) if you want to mandate that high tech manufacturing come back to America (e.g. "just make iPhones here" which seems to be a common sentiment in my circles), it would be foolish to suppress an industry creating insatiable demand for high tech components and ensuring that it goes offshore at a time when other countries are also trying to build more local manufacturing. To say nothing of components and construction materials that are already manufactured here. (2) there are very few types of industry that would create less local environmental impact than a data center, no chance if you think data centers use too much water you'd be okay with the toxic chemicals that chip fabs work with. (3) since America is a high wage country with a lot of R&D strength, any factories we build are naturally going to be much more automated than they are in low wage countries. Being against entire industries because individual facilities don't create enough jobs would probably be quite limiting in the types of manufacturing you'd approve of as well.
I have several friends who used to lament the loss of manufacturing jobs as a ticket to the middle class, but now say they're going to protest a proposed data center, which feels a bit ironic. None really link it to AI's social impact like Gizmodo does here, the argument always starts with "I don't understand what they need a data center for" (often genuinely wanting me to explain it since I work with computers) and then goes into noise, water use, or loss of farmland. I'd probably not want to live near the noise pollution of a data center or any other kind of noisy industry either, so their views aren't incomprehensible or anything (though the farmland one makes zero sense to me), but it does seem like an instance of the revealed preference that many Americans are just deeply skeptical of anything more intensive than an Amazon warehouse going on in their area, even if they enjoy a fantasy version of the country where (usually other) people have a nice union job in a widget factory. It's good to remember when political extremists try to claim there's some easy fix that will make America an industrial powerhouse again; in reality, most of us don't want anything close to that.
I'm generally defensive of Reuters coverage of Tesla (most of the things Musk has called lies turned out to be true) but this one does seem overly sensationalized. About half of the article focuses on the fact that Tesla does precision mapping and annotation of routes...which yes, all self-driving cars rely on this, it's a major safety advantage they have over human drivers that don't have access to a 3D map and near real time feed of road conditions everywhere. They also have teams scrutinizing collisions and near misses to improve the system. I'm not sure why Reuters frames all this as a deceptive "behind the curtain" thing or a strike against the safety of the system.
The rest is about the flaws of Tesla's safety statistics, which Tesla themselves appeared to acknowledge on their last big earnings call where they said wider launch of Robotaxi would have to wait for software improvements to improve safety. This is why getting safety information from a neutral party is so important.
Implicit in a lot of "AI jobs apocalypse" predictions is the assumption that most tasks are ridiculously easy compared to AI research, so naturally the smart AI researchers can understand any profession well enough to credibly predict that AI will be able to replace it. I'm personally not sure the apocalypse has been truly disproven as opposed to progress just being slower than some of the overexuberant predictions, but there does seem to be a pattern of famous AI researchers predicting a job would be automated and turning out to be wrong because they focused too hard on a single aspect of it that could be automated while handwaving or ignoring the hard parts. This has prominently happened with radiology, then with customer service, and now they are walking back on programming too. Maybe take these guys with a grain of salt going forward? I trust them to be able to tell us frontier AI models will keep getting better, not to predict the impact that will have on specific industries. Some people will insist we should give them half credit for predicting there would be impact at all (as opposed to the "it's a bubble" refrain) but I think it should be possible to ignore two categories of obviously dumb predictions at the same time.
I think the most important criteria with a reader (aside from hardware quality) is whether you're comfortable going outside the manufacturer's store to buy DRM-free books, or at least ones that can be liberated from DRM for future proofing. Calibre still speaks the format of these old Kindles, so they're usable, I expect that will continue to be the case for Kindles. If format conversion is too annoying to deal with then it's better to read on a general purpose iOS or Android tablet. I have a Boox NA4C and it's ok, nice hardware, but I have noticed the constant phoning home and am annoyed by the GPL issues (not that I expected a Chinese Android device maker to be fulfilling their open source obligations). For that reason and others I've mostly come around to just reading on a phone and tablet with non-eink screens.
I used it for about a week, thought it was an interesting demo of the possibilities of general purpose automation with a local model (even though most OpenClaw users use hosted models). The approach to scheduled jobs still makes more sense than anything else I've seen implemented. But like a lot of self-hosted software with passionate evangelists, it wants to be your new main hobby instead of just getting out of the way, and I lost interest because I didn't want a new hobby. It feels like a more thoughtful community could have made something useful with the concept, but as it is the community around it is too absorbed in marketing and shipping stuff for its own sake.
Good quote from the author's earlier post about iCloud Photos:
> Software and services need a warranty. Until they have one, we completely control how much we value our data. That is the best we can do.
Best to treat these photo sharing apps, commercial or open source, as social media. Would you use Instagram or Flickr to store your most important photos and delete your own copies? I would not, same applies to Apple/Google Photos and similar apps. Besides the risk of the company suddenly shutting down or (more realistically for big tech) changing how their service works in a way that makes it useless to you, even if self hosted it just adds a bunch of things that could go wrong which don't apply to keeping it in a folder somewhere with an offsite backup. Filesystems don't have a warranty either, but at least they're easier to reason about.
As I understand it, "mercenary spyware" is Apple's preferred euphemism for the "(semi)private israeli companies selling their solutions happily to all regimes regardless of consequences"
I was agreeing with kvuj and rguyorama that the original link is to an announcement that an investigation is happening, and it's too early in the process to productively discuss it. People have very strong and emotional pro or anti stances on the Tesla Vision system in general, and love an excuse to have the debate again, but in the comments here where people are talking about their stance you might notice that they don't reference any specific facts from the linked report to support their arguments. This is because the report is still vague at this stage and doesn't provide any specifics that inform the discussion.
You’re just reciting your priors, which I think supports GP’s point: no one is getting new information out of the posted link, so it’s probably premature to comment on it.
During the Biden administration there was a whole campaign to try and get Wikipedia to recognize the recession that had been declared by Fox News pundits. The liberals are characteristically more creative with their version, but it still sounds like partisan wishful thinking (awfully nihilistic, too). One could slice and dice the numbers any number of ways and it could fool a layman like me no problem. The best defense I know of is to ignore any analysis that tries to change the definition of a recession.
For the purposes of the discussion at hand, yes some results do ultimately come from Google, just via third-party SERP providers rather than Kagi paying Google for access since Google doesn't offer their own public API (and neither does Bing anymore).
Whatever you think of the ethics of doing this, it does hurt the reputation of the follower labs in my mind. If their capabilities can't exist without the work of the frontier labs, they're less equal competitors and more the guys trying to sell you a shoddy knockoff. Not that there's no use case for shoddy knockoffs.
With the disclaimer that I haven't tried to set up any kind of agent-to-agent messaging so it may be obvious to those who have, what's the reason I would want something like this rather than just letting agents communicate over some existing messaging protocol that has a CLI (like, I don't know, GPG email)?
>They are also addicted to the gambling mechanics baked into these LLM powered tool's UX. "If I write this prompt this way, I'll get better results" is the equivalent of a gambler being superstitious about how people behave while the cards are being dealt, or in which order they press the buttons on a slot machine.
I realize this feels good to write and that's why people say it, but I can't help chuckling at seeing it combined with "stochastic parrot" in the same comment since the two descriptions are mutually exclusive...
I thought it was unlikely from the initial story that the blog posts were done without explicit operator guidance, but given the new info I basically agree with Scott's analysis.
The purported soul doc is a painful read. Be nicer to your bots, people! Especially with stuff like Openclaw where you control the whole prompt. Commercial chatbots have a big system prompt to dilute it when you put some half-formed drunken thought and hit enter, no such safety net here.
>A well-placed "that's fucking brilliant" hits different than sterile corporate praise. Don't force it. Don't overdo it. But if a situation calls for a "holy shit" — say holy shit.
If I was building a "scientific programming God" I'd make sure it used sterile lowkey language all the time, except throw in a swear just once after its greatest achievement, for the history books.