Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Our First Generalist Policy

physicalintelligence.company

Post reply on HN
Print view View original

131–140 of 249 posts

Re: Our First Generalist Policy

#131

Post by ben_w »

p1esk wrote: post #96
bitmasher9 wrote: post #89

Earlier quoted context omitted.

Bifurcation of society with fewer and fewer people moving upward in social status. The poor have nothing. The bar of assets required to not become poor continues to raise. For example, let’s say you have 5M NW, and 75% of it is in an uninsurable residential real estate. Your house is at high risk of being destroyed, and if it does you barely have the assets required to rebuild. If this happens twice you are have noth…

Let’s say I have 10M in stocks. What is the optimal strategy to not end up as “have nothing poor” in an automated economy?

Make sure the stocks that you own include some of the companies making the robots and/or the AI that do the be automation.

And/or, if possible, buy yourself one small Von Neumann replicator.

Re: Our First Generalist Policy

#132

Post by ben_w »

forgetfreeman wrote: post #113
prettyStandard wrote: post #23

Earlier quoted context omitted.

No health problems if there's no people to have them.

Additionally issues around global carbon footprints quickly become tractable when you eliminate the bulk of humanity.

Only at the 99.9% level, and only briefly before the majority of the remainder look at the open and empty planet and go "oo, nice, free land for a big family".

If you "solve" this in any way other than giving people better options for them that are inherently also better for the environment, it's unstable — even if almost every nation tries to enforce it at the same time, whoever defects (in the Nash game theory sense) literally inherits the Earth.

(And that's why I'm also expecting Von Neumann machines to be the environmental disaster of the solar system within a century of someone making one: game theory says that whoever does that sucessfully, inherits the future light cone).

Re: Our First Generalist Policy

#133

Post by closewith »

TylerE wrote: post #120
erulabs wrote: post #99

It almost feels easier to disassemble and resew the shirt from recycled fabric. I’m mostly joking, but my point is that physical AI probably implies a complete rethink of every individual routine from first principles: why fold the shirt at all? Why not just-in-time-ironing? We’re focusing on the hard problems because we’re imitating how a human with limited resources would approach the issue. If you asked a robot to…

That could just as easily be turned on it's head too - if you don't need skilled (or any) staff to shop, cook, serve, and then washup, why would you ever NOT eat at home? Especially if it can operate very quietly, one fairly slow robot could do probably all the housework, and could do it at night when it's literally out of sight and out of mind. It would feel like magic . You'd wake up every morning to a clean house…

Most people go to restaurants because of the social benefits, not the labour savings.

The idea of people or groups of people siloing themselves into their ultra-convenient homes and never interact with others is a dystopia and a sure sign of an already broken community.

Re: Our First Generalist Policy

#134

Post by forgetfreeman »

ben_w wrote: post #132
forgetfreeman wrote: post #113

Earlier quoted context omitted.

Additionally issues around global carbon footprints quickly become tractable when you eliminate the bulk of humanity.

Only at the 99.9% level, and only briefly before the majority of the remainder look at the open and empty planet and go "oo, nice, free land for a big family". If you "solve" this in any way other than giving people better options for them that are inherently also better for the environment , it's unstable — even if almost every nation tries to enforce it at the same time, whoever defects (in the Nash game theory sen…

I must have deadpanned that a little too hard. Was intended as sarcasm.

Re: Our First Generalist Policy

#135

Post by ben_w »

forgetfreeman wrote: post #134
ben_w wrote: post #132

Earlier quoted context omitted.

Only at the 99.9% level, and only briefly before the majority of the remainder look at the open and empty planet and go "oo, nice, free land for a big family". If you "solve" this in any way other than giving people better options for them that are inherently also better for the environment , it's unstable — even if almost every nation tries to enforce it at the same time, whoever defects (in the Nash game theory sen…

I must have deadpanned that a little too hard. Was intended as sarcasm.

Ah, fair enough.

Difficult to tell, as I've seen that position held seriously here.

Re: Our First Generalist Policy

#136

Post by nooumenon »

lachyg wrote: post #5

(I work at π.) Happy to answer any questions on the model, hardware, etc

Hi! Very cool results. Are you able to share some numbers about the slope of the scaling curve you found, i.e. how performance responds to a growing nr of demonstrations?

Academically I'd also be very interested how much of a data efficiency improvement you achieved with the pretrained model + task specific post-training versus from-scratch task specific training - like, if post training requires say 50 additional demos, and from-scratch on smaller model requires say 250 demos (or whatever) to match performance, that would be an interesting quntification of the efficiency benefit of using the big foundation model

Re: Our First Generalist Policy

#137

Post by someothherguyy »

bitmasher9 wrote: post #89
p1esk wrote: post #84

Earlier quoted context omitted.

What will happen to those of us who are not rich but also not poor? Upper middle class with net worth between 1 and 10 million?

Bifurcation of society with fewer and fewer people moving upward in social status. The poor have nothing. The bar of assets required to not become poor continues to raise. For example, let’s say you have 5M NW, and 75% of it is in an uninsurable residential real estate. Your house is at high risk of being destroyed, and if it does you barely have the assets required to rebuild. If this happens twice you are have noth…

> Your house is at high risk of being destroyed, and if it does you barely have the assets required to rebuild.

Or it becomes worthless when no one else has anything and demand tanks

Re: Our First Generalist Policy

#138

Post by soco »

nightski wrote: post #76
carapace wrote: post #70

Earlier quoted context omitted.

> If the technology really advances to that point it will be very cheap to live. Yes. But the flipside of that is that it's hard to earn money. The cheaper it is to live, the harder it is to make money. The problem UBI solves is when most humans can no longer compete with machines in the economic system. Once that happens there are only a few options, and the least unpleasant is simply to give them (us) money tokens.

Humans don't need to compete with machines in the economic system. Machines aren't out to make a profit. Only humans are. The truth is if this technology really advances to the point we are talking about money won't really be a relevant concept. Money at it's essence imho is a claim on future labor. If that labor is extremely cheap, then all money is nearly worthless. But LLMs/transformers are not this technology. We…

Machines aren't out by themselves for anything. The machine owners on the other hand, they are out to make a profit. So human workers compete with machine workers for the profit f the owners, and at least here I'm sure we can agree who wins in the end. Because as popular the comparison with the horses vs cars might be, at some point the machines will be smarter than humans with or without AGI. From that point on, any job a human would be able to do, and I mean any job invented or not, a machine would do cheaper and probably better. Zero options for the jobless humans from that point on, zero money to spend except welfare/UBI. And even that doesn't sound very appealing, because humans will look at the machine owners with jealousy and yeah I expect quite some fuss with many only unpleasant outcomes.

Re: Our First Generalist Policy

#139

Post by golol »

forgetfreeman wrote: post #115
golol wrote: post #26

Earlier quoted context omitted.

Inbetween the current world full of labor scarcity, and the philosophical dilemma "what do I even do" post-scarcity utopia, is a world similar to our current one with much less labor scarcity and much more quality of life. That's what we're aiming for right now. What comes afterwards we can worry about then.

Quality of life for who, though? That's kind of the point of all this back and forth isn't it? Because if prior examples are anything to go on the likeliest outcome is we're talking about further increasing QoL for a small minority of individuals who least need it at the expense of basically everyone else.

The ever expanding middle class of course.

Re: Our First Generalist Policy

#140

Post by Philpax »

forgetfreeman wrote: post #114
torlok wrote: post #94

Earlier quoted context omitted.

Fair, but how will robots change that?

Ostensibly there won't be any carpenters to hire because they've all been bankrupted by automated contractor services. This seems like a pretty big change to me.

But how does that change the calculation in the grandposter's reply? You can still do it yourself if you want, or you can hire the automated contractor if you don't have the time or effort.
Post reply on HN