LLMs already solved the "System 2" part of this, to borrow from Kahneman, it's the "System 1" part that's lagging behind here. Current Claude/Gemini/ChatGPT is more than enough to tell a robot what chores to do, what to do with a thing, how, where to put it, etc. but what's still missing is the ability to reliably translate those goals to movements of a robot in diverse and tight environment that is a typical house or apartment, with any kind of reliability and safety.
That one has been a well-known thing for a decade if not more; it's not just Google, half the Internet will start throwing captchas or denying access once you connect via a VPN (specifically "VPN" as in one of the services you pay to avoid location-based discrimination of media streaming platforms).
> I feel like in this case, the “Amazon or Netflix” example is particularly bad because I feel like I’m actually drowning in streaming providers.
It's a fake choice, because they carry mutually exclusive catalogs, and entertainment choice is not particularly substitutable (e.g. if I want to watch "Star Wars" and it's not available on services I'm subscribed to, I'm not going to be satisfied with all the rich selection of things they carry that is still not "Star Wars").
Lots of that in the economy, that's where the most money seems to be made. Smartphones are my go-to example: plenty of nearly identical options to choose from, choice entirely set up by vendors, with little to no way of users to voice their feedback. A supply-driven market. You get to choose from what's made available, not what is possible.
It doesn't feel like a fake choice when I sit down on the couch and flip between 5 services trying to find something to watch, and on a weekly basis I will watch something on almost all the services. The content nowadays is generally fine, and it is difficult to pick something when everything is pretty close, and every now and then there is something really good, so maybe competition is working here to raise the bar for all services? Not sure.
Saying entertainment is not substitutable is kind of crazy. It is very rare that I will only ever want to watch one specific thing, and if that happens, I have the choice to rent it, or buy it or pay for one month of a service. To me as a consumer...that great.
The market for smartphones is mostly driven by lack of options in OS, not hardware. Each of the big players offer plenty of hardware choices at different price points. But if you don't like the OS, harder to get away from that. Competing on OS is very difficult though.
Every day when you lower your butt onto your chair, you trust a stochastic system enough to assume you'll rest on the chair safely and not spontaneously phase through, which would lead to rather gory and painful terminal experience.
Physics at macro scale is stochastic, which is a good reminder that stochastic != uniformly random. Expected distributions matter.
> QM has such small standard deviations as to be irrelevant on the macro for things like bums and chairs
I was going to segue into thermodynamics as a backup example, but you made me think of something better.
> IMO a better example would be the stochastic nature of quality control in manufacturing.
How about, more specifically, food manufacturing? Or maybe, let's talk about cooking?
Cooking is as stochastic as it gets, and we handle it fine. It could be better - the better version is called "chemical process engineering", it's what cooking looks like when you care about quality and consistency of output, and can afford the equipment and process actually necessary for it. Regular people don't (i.e. neither care, nor can afford) - we call this cooking. It's an art, not a science, and people not only do it, but love it, and tie their identities to it, and build businesses around it, and a culture that embraces all the compromises (and calls the more serious approach "unhealthy").
There are too many value judgments in this post. You can "cook" like "regular people" do, and be completely serious, and apply chemical and physical knowledge in doing so, and test the output for quality; generally that's what restaurant chefs do. It doesn't make sense to cook like you're tooling an assembly line, because you aren't cost-optimizing and packaging a product that needs to sit on a store shelf for weeks, months, or years while maintaining its desired qualities.
Speaking generally, food produced though "chemical process engineering" (a.k.a. factories) must compromise on many axes, one of them being nutritional content. We intuitively do not care about several of these dimensions when cooking food with fresh ingredients, at least not at the scale of, say, Kellogg's or General Mills.
Maybe that's evidence of accepting a stochastic process in our daily lives, but you're kind of selling the tradition and science of cooking short when you argue that factory-produced food is a "more serious approach".
> Cooking is as stochastic as it gets, and we handle it fine. It could be better
My attempts at making bread have been too stochastic, in that it hardly ever produces nice results.
But yes. Eyeballing how much dried herbs to put in my dishes because I like what 2-isopropyl-5-methylphenol does for them. Usually it works, sometimes it's just a bit too Italian.
Might not be the amount, you may have not controlled for humidity or temperature (wink wink), or just that the timer on your oven is off by one minute per every ten minutes, and its bang-bang thermostat never actually reaches the temperature you set on the panel, and...
... in some sense, it's a miracle most people deal with this kind of bullshit without complaining much.
(Probably because they don't realize it's something to complain about. It's just how things are.)
Personally I don't see this as joking, I think this whole space is severely underfunded and could use some publicity and moonshot contests. I mean, think of it, the planet Earth is full of beautiful and diverse nanotechnology that can literally map-reduce complex behavior over individual molecules, and we do so little to use it for practical purposes. Even most advanced manufacturing methods we use are still simple things applied in bulk, counting matter by volume instead of as objects. There's lots of unexplored potential within reach, and here we actually know it can pan out, because we see these processes happening everywhere, all the time, all at once, all around us.
Maybe "edge case engineering" is precisely what they're looking for? Get people to think about beating the rules with cheesy strategies, in hopes some of those could, with some cleverness, scale up and evolve into proper, broad-range solution - or at least become a key previously-missing component of one. But even if it can't, very narrow capabilities can still be useful too; military isn't beyond doing silly things if they offer enough tactical advantage (enough to offset extra burden on logistics, at least).
That's your prerogative, but be aware you'll continue to remain confused about LLMs. Anthropomorphizing them is what gives you the best high-level intuition about where and how to employ them, and where and how not to.
reply