Hacker Newsnew | past | comments | ask | show | jobs | submit | OroPla's commentslogin

Built once, all they need is energy. They can work 24/7 365 days of the year.

I do not think humans can compete on price, for one robot will be worth at least three humans in hours worked per day. Question is only if such a robot can be made and how long it lasts until ceasing to function.


Built once, they then need periodic maintenance and replacement like any mechanical device (general-purpose robots will be exposed to real-world stresses and danger, after all), and then there is the question of whether the cost of these services outweighs the cost of human labour. In countries where human labour is extremely cheap, e.g. the Indian Subcontinent and GCC countries, it may take a long time for the cost advantage to decisively shift towards robots.

They will repair themselves.

I was intending to snark something like "With pieces that they build themselves from minerals that they mine themselves in country that they rule themselves ?"

But then, as often in those debates, it's hard to say something ludicrous enough for the answer not be a terrifying : "well, yes ?"

Anyway, I hope all the new anthropic millionaires will enjoy life while it lasts. A couple of them might fund companies that will help us survive a few more years. But surely the IPO was worth it, right ?


Consider that in the US, internal combustion engine personal cars have been thriving industry from the 40s to... arguably today. And yes, they need periodic maintenance, replacement, and some uhh... simplification of the environment, and yet all they do is transport the labor for a few hours a day, sitting idle most of the time.

Most households own a car which even if you go used has a price tag of around $20k if you want something average. It wouldn't be much of a stretch to see a future where every household owns a $20k AI computer. Or two. If you are fancy you own a $100k AI computer.

However, since there is an ongoing war on general computing it wouldn't surprise me if instead people get locked into subscriptions.


Shit people be telling ChatGPT isn't protected by law. You get into any trouble, your chat history gonna be gone over with a fine tooth comb. Some of the masses gonna realize that and gonna want a sever rack at home instead.

I bought a 2000 Honda Accord V6 from a friend for $3k a long time ago. I drove it for several years, adding probably another 90,000mi to it. It didn't need much maintenance during that. It's still kicking from what I understand. An incredible car, I'm a little sad I parted with it.

I still think of the economic benefit I got from that car. It drove me to university classes. It drove me to work. It took me to see a total solar eclipse. It brought me to see distant family regularly. I went on a lot of road trips on it. For roughly $3k of capital costs.

I look at most expensive things and size them up as "how many Accords is this?"

Almost 7 Accords is a lot.


Yeah, I remember that last decade the prediction was the 2040s, but then LLMs happened and it became 2030s.

It's misguided, but it works reliably on every human. If you are not losing weight you haven't reduced calories enough. If you don't lose weight on zero calories for a year you are lying.

Also, (simplified) mass is a form of energy, so I am not following how that assessment would explain the fault with the approach.


Why not jump to a more simple theory of mass in/mass out then? Why care about calories at all, especially as they're not a biological unit either ?

Trying to actually and reliably measure what happens on food intake at an individual level is near impossible anyway, so keeping the explanation simple would be a step in some direction IMHO.


Because drinking a kilogram of melted ice cream is not the same as drinking a kilogram of water. One of them has zero calories, the other does not.

Eating 100 calories of paper and 100 calories of sugar hasn't the same effect either.

Neither units make sense in actual practical terms.


Paper doesn't have calories in the context of human diet.

Calories are fine and I don't understand your objection. If you would rather use joules, some countries do that and it doesn't make any difference in practice.


A calorie is a unit of how much heat is generated by burning something, in principle it has nothing to do with human diet.

That's my objection, and I look at the contorsions diet people do to justify talking in calories the same way we talked about bad smells making people ill 6 or 7 centuries ago.

I don't think we should stick to something fundamentally wrong just because it has been in the zeitgeist for a long time.


The calories on food labels are adjusted to roughly match digestion. They're pretty good. And you haven't explained what you think is better.

Or to put it another way, they are a biological unit, not raw numbers from a calorimeter.


People can drive a car without understanding it deeply enough to be able to troubleshoot and repair it. But few drivers are under the delusion that they are car mechanics.

You don't need understanding to use something. For example, I have a grasp on both sarcasm and irony. I recognize it when used, but maybe not always. I can use both myself. But I have trouble explaining the concept of irony, since I do not truly understand it. Sarcasm on the other hand is easy to explain. It's mocking someone with nice words.


A lot of work went into transformation of the car into something that is easy to use and easy to explain to a degree that makes it usable. Much of the complexity is hidden behind simple interfaces. This happens everywhere to provide simple building blocks so people in other domains can use them. Often it doesn't come for free but usually it's worth it.

I can't explain these since I don't understand them. I can at most regurgitate some points about each that are true.

If you truly understand something, I don't think it would pose a problem to explain it to a three year old using only vocabulary that the three year old knows.

Realizing you can't do that is realizing you do not truly understand something.


You clearly have no experience with small children. Until the age of 10 or so, they can mainly grasp concrete ideas. Take the famous Piaget experiment. You can explain height vs volume to them in their terms, but they'll still reach for the tallest glass. At the age of 3, they cannot understand NP, Heartbleed, relativity or Monty Hall, no matter how it's explained. If that were true, we would all have graduated age 6.

There's a difference between explaining and teaching. Someone smarter than me can explain to me how quantum physics works, but that doesn't mean I will understand how quantum physics works afterwards.

If explaining was all it takes to acquire knowledge, everyone leaving school would have a deep understanding about every subject.


Then the title of this article doesn't mean anything.

Isn't delegating the unfun parts to focus on the fun parts exactly what we wanted from AI?

I don't like writing a ten pages report, nor do I like reading a ten pages report. If it was up to me we would just communicate as tersely as possible everywhere instead of filling things up with a lot of nothing. It's why I rarely read (or write) articles, but read a lot of headlines and comments instead. One gets to the point faster this way.


what if the fun part is writing code and the boring part is figuring out requirements? that's me right there: out of a job

If AI can solve your problem, why should anyone else spend their time on you? It's the new "just google it" and it means the person you are asking doesn't care about you.

If someone is being lazy and is somehow too lazy to even type their knowledge recall query into Claude Code, it may be reasonable to refuse to help, but there is a productive, non-hostile way to do it.

Every time someone asks you for help you should ask what they've tried so far so that you don't try the same things.

If they say they've tried nothing and put no thought into it, and haven't even thought about possible approaches that is pretty disrespectful of my time.

Even so! They may not realize this and there is an empathetic, polite way to let them know this the first few times. Getting oneself unstuck without becoming an agent someone forgot to set /goal on is a skill that people sometimes need feedback to learn.


I'm not asking them for knowledge based stuff. I'm mostly asking them for how the team does things in terms of software development processes / pipeline after switching teams, for example.

> why should anyone else spend their time on you?

This is an extremely unhealthy mindset. Asking questions is part of socializing and can be valuable even if you could just google something.


I don't subscribe to the idea that there ever was a reliable source of truth. It's just that having open access to information makes it so people see competing versions of framings some of which are incompatible with each other. For example, Christianity, Judaism and Islam can't all be true. They could all be false, however. Same applies to anything else. Just because you clearly know that this other view is wrong doesn't make your own perspective right.

Unfortunately or maybe fortunately it is entirely possible to live a long and happy life being entirely wrong about a great many important things without realizing.

I can't be sure that what I think is true, is really true. Therefore I don't really want to make the decision on how other people are supposed to live their life and find anyone that thinks himself worthy of doing just that suspect. I prefer someone that exploits people for their own gain, since then we are at least clear about the incentive structure and can measure and verify that gain and also negotiate for a cut. Something that isn't really possible when we are told to do things for our own good because God said so.


Maybe it's worthwhile to remember that money and paying for things arose from the necessity of trade. Different humans created different value and traded with each other to all benefit. To not have to trade goods, money came to be.

If your AI can give you the benefit without having to trade, you no longer need money. But even if your AI can do everything a human can, unless it can create something out of nothing, you still need land. Both to grow food and to extract resources. And you still need to defend your land.


> Maybe it's worthwhile to remember that money and paying for things arose from the necessity of trade.

Initially, yes. But that was in a situation where (broadly speaking) people had roughly equal levels of resources, and they developed money to have a consistent way to value those resources.

Here and now, the money the super-wealthy have now represents control of those resources. You and I don't have much to offer in that context.

And if (as CodesInChaos alludes to) we are positing a future where the super-wealthy controllers of AI also have sufficient robot-power to replace human manual labor, why would they not also have sufficient robot-power to kill anyone trying to take those resources back from them? Certainly, this is the calculation many of them are already making.

It is abundantly clear that if they are allowed to continue down this road, beyond a certain point the only way back will be by overwhelming force. And if they are able to continue farther, then beyond a certain point assembling that level of force that may no longer be practical even in the rosiest of scenarios.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: