dogcomplex

Wait a few more months... let's see what the bittorrent and private crypto folks end up doing with AI... Either way, IP is going to either accept defeat or get very very loose in its claims, as the ability is almost here to just process a feature-length film, slightly modify each scene and replace script, camera angles, environments and actors with generic IP-free amalgamations, all while still capturing the tone and story... Images/text/code are already done for. Who's going to enforce that, and how much similarity do you need to infringe? Who's going to prove you processed that film when nothing about your movie is quite the same?

Entirely open (at best) question, which no smug hacker news commenter is going to answer. What does one do?

(My current longshot hope? Digital democracy, on the backs of open source production economies running on the latest AI for highly-accessible/affordable data processing / labor. If we just start making online group decisions and scale it up, that's a power bloc that can run its own parties and strongarm existing gov processes - assuming the network even wants to interact with them... )

I hear ya bud, and I think you have the right take. This will ultimately vastly improve the lives of quite a lot more people than if it never happened. If we survive the coming years of turmoil and AIs or social unrest dont destroy the world, we are looking at post-scarcity future of so much plenty that the powers that be would have to do something particularly obviously evil and nefarious to keep that wealth from spreading to everyone. As the cost of mass production of basic needs, housing, etc etc all drastically lower, the excuse for why they're not being spread gets more and more ridiculous - even as people's individual power lowers as they become redundant. (Though reminder: the current chess champion is a human with AI assistance, not a pure AI. Humans copilots may still be semi-relevant a while longer!) Wealth Inequality might very-well still skyrocket, but overall this should still uplift everyone's quality of life and effectively allow for mass retirement - even in the third world, eventually.

IF we make it through...

And, crucially, the entire project is preplanned. That preplanning is the software phase equivalent - the construction is running the code. (Or at least, the preplan is defining all the necessary function definitions and connecting them confidently, and construction is just writing each implementation.)

Ask a toddler how the world works and you'll get a very similar response. It is entirely likely the 80%-of-human-intelligence barrier is not a "dead end" but merely a temporary limitation until these models are made to hone their understanding and update over time (i.e. get feedback) instead of going for zero-shot perfection. The GPT models incorporating video should start developing this "memory" naturally as they incorporate temporal coherence (time) into the model.

The fact we got this far through brute force is just insanely telling. This is a natural phenomena we're stumbling upon, not something crafted by humans.

Also - fun fact, the Facebook Llama model that fits on a Raspberry Pi and is almost as good as GPT3? Also basically brute force. They just trained it a lot longer and it shrunk the model. Food for thought.

It is very obvious there is a mass unemployment wave coming - or at least a mass "retraining" wave, though the new jobs "teaching AIs" or whatever remain to be seen. I hope everyone currently just questioning whether this will happen now is prepared to state it with conviction in the coming months and fight for some sort of social protection program for all these displaced people, because the profits from this new world aren't getting distributed without a fight.

Replying to this to remind myself to research later, as this is a very important question

Yep, exactly. I hope and expect to see automated vertical farm systems as cheap consumer products soon enough, for one. Though I also don't think we necessarily need to reinvent the wheel with this stuff - it doesn't have to be some independent micro system (which are typically much less convenient than mass-produced systems). We can still use standard farming/water/energy-production/sanitation - we just need to reduce the bottom-line cost of those services (which I am still strongly betting AI will impact hard) and make it so creating a competing system takes much less capital to boot up, in case existing services decide they want to try and charge the same consumer prices while eating up all the efficiency gains from AI.

We don't need to settle for lower quality, or convenience. We just need to ensure capitalism does what it's supposed to, and pass on the efficiency gains down to the consumer.

Though also, for my money - water, energy, sanitation? Those are already very cheap, and will get cheaper naturally. And food is cheap if you keep to basics. Automated housing construction and automated high-quality-meal kitchens (not just basics) seem like the areas to tackle for real cost savings. Ultimately the last bastion will be rent seeking on land though - which could be alleviated by creating housing alternatives off the conventional grid, or uplifting the quality of services in much of the developing world so there are much more options of safe areas to live (which again, is happening fast already and will happen much faster with a rapid improvement from AI tooling).

My argument is this is the natural way capitalism works - or is supposed to work - and we should treat that with hearty skepticism. But the only way this doesn't happen seems to be if there are monopolistic forces actively holding it back. Otherwise there really should be an abundance of cheap options for living coming to consumers in the near future, which would effectively be close to a UBI, and a "living off the land" alternative to wage labor.

(P.S. for housing, my bets are more on modular housing production in a factory setting becoming highly automated - the quality level of those builds is already quite high. But also just automation on the materials production all the way down the stack. There's hope for in-place construction savings (robots!) but I think that'll come much later. Logistics transportation was already due for some major improvements before AI reared its head - cheap electric engines just enable some very different structures that effectively have a net-0 cost of operation when combined with solar. Any innovation on that front would dramatically reduce the cost of everything else.

Dark kitchens are already a thing, and expect them to keep growing - especially as the tools improve. You could probably supply a whole city with the same wide variety of well-cooked dishes from a single facility with efficient transport. There's an element of dystopia there as cook jobs disappear and thus the personal touch, but the consumer reality (should be) basically-free meals at the same quality we expect from restaurants, on demand.)

Due to the nature of poor people having no economic option except to do labor, the above system would effectively only return the money back to the workers. Anyone with enough wealth accumulated to no longer need to work would be paying more in taxes than they receive by UBI

As for whether the entire system of investing materials/capital when a project is risky should be rewarded at all (ostensibly the rich class' job) - mine is a conservative take. It could be rewarded less, certainly (via a wealth tax). Completely removing that mechanism ("money illusion") seems like a recipe for a society where there is little point in taking any risky venture.

But the real question is: will we make good pets?

It's not necessarily extraction from those that do work for a living if the tax to pay for the UBI is largely a wealth-tax taking the majority from wealthy investors (who, by the conventional wisdom of capitalism, do exactly what you accuse and extract time and output from those that do work for a living for the "hard labor" of managing an index fund). If that tax is balanced to offset the two, well - you've got a fair society (and one likely to eliminate billionaires over time).

I estimate that wealth tax price to be around the average cost of living - as any less and those are people who have no choice but to accept whatever poor labor deals are put in front of them, which enables the capitalist to extract more value than the labor is worth. If people have their costs of living covered and freedom to pick and choose to labor or not, then it doesn't matter if your net worth is 10k or 10M, you're going to accept a similar wage for similar work. (well, not entirely - but at least at that point it's about desire for luxuries not hard desperation)

A truly Communist economy takes this much further and splits things evenly from the start. Though I'd argue you get there with a very-high wealth tax and UBI. That doesn't necessarily mean the society is producing a lot of goods though - especially if you just uplifted most of your citizenry that were only doing those shitty jobs because they would literally starve otherwise. But to communism's not-credit, there haven't been many great implementations solving the motivation problem yet either.

I suspect a gradual UBI/wealth tax combo is going to do a lot better at easing in motivation and crucially - preventing violent backlash from those in power - than any sudden shift to a fully-communist system. Personally, that is.

And the purely-digital "robot slave" will likely be $0 (albeit probably a little lower quality than the premium corporate versions), if open source software patterns are any indication.

Too limited scope. It's entirely possible increased profitability will not create jobs if it outperforms most people.

However, reduced costs and increased profitability do mean increased/easier quality of life. Even if robots are outperforming people in the business sphere, you're still gonna be seeing the externalities of incredible abundance at the local level. Hard to say how that wins out on net when people can't get jobs to pay for the (very cheap) goods and services though. This is typically where government/community should step in.

With the followup break down "Will AI decrease consumer prices on goods and services?" and "Will the balance of overall quality of life be negatively affected?" since simply focusing on jobs themselves without the externalities is too limited. ("Yes", and "Hard-No-in-long-term / Possibly-in-short-term")

> I do not think our current social policies are ready for that.

Understatement. (And also, they weren't really ready for the previous waves either)

The upside is we exist as both producers and consumers - and while our value is about to go down (or be voided entirely), so is the price of all the things we want/need. Can't afford a house because the Atlas robot outperformed you at construction? Well, get a few buddies to pool funds to buy an Atlas robot yourselves and get it to build your houses. The barrier of entry is similarly dropping fast. It's gonna be a weird, sad and awkward short-to-medium term as the fluctuation of which drops first - employment or consumer prices - plays out, especially if government doesn't step in to protect people. But long-term I think it's hard to argue that there's not going to be an enormous increase in overall quality of life. (barring some sort of annihilation event...)

With AI-assisted production tooling, a lack of a UBI in the coming years should be basically treated as an act of war. The bar to achieve it is going to be ridiculously low - the only way it's not happening is if the powers-that-be are actively preventing it. At that point any mob could and should just spin up their own AI-powered factories for basic needs regardless of corrupted government/big business approval.

Quite possibly. Though if you have that many cheap knowledge workers looking for work, that's quite a lot of brainpower to be dumped into the few remaining unsolved problems in robotics mass production (not that there are many - those robot dancing videos are terrifying, and Cobot arms are getting extremely versatile and affordable). i.e. digital work might be the first to go - but manual would follow quite quickly.

I would very-much love to see that chart with relative prices from other countries, with taxation accounted for in pricing.

Company size drops too though. Enough automation and every person has the abilities that took hundreds of people in their parents' day. Sure, the wealthiest will be able to leverage their gains the furthest - that's just how capital works - but it's also reducing people's dependency on these big firms and enabling them to personally capture the gains from the drastically-improved means of production. If and when communities are able to self-sufficiently produce most of what they need with just a few people and some AI tools, the game changes. Sure, at that point the rich would already be building death stars and wealth inequality would be even more laughable than it is today, but it'd at least still be a step-up in quality of life from the current regime.

If "interesting" is the keyword here, I'd have hope that language models based around mathematics research would at least be able to narrow the scope of theorems being iterated through to those closest to what mathematicians are already talking about. We'll see

This is the most cynical take, but also sadly the most true to history and therefore probably the future. Sure, the future might have every one of us with the ability to boot up with a command a self-assembling automated factory that previously took a supply chain of thousands of workers to run, but due to the suddenly enormous cost of land and raw materials relative to our make-work wages, nobody can afford to build it.

I am somewhat serious, but also being a little factitious here. There's a breaking point of widespread democratized (runnable by like 1-5 people, or a small community) technology where these companies don't have the same ability to absorb all these gains. If/when the basics for living are largely automated and able to be bootstrapped for very cheap / very little labor by AI-assisted tools, then the only way people are going to settle for jobs that give less than what these provide is by top-down dictatorial artificial scarcity and rent-seeking, blatantly destroying this production ability. I think there's definitely a cynical reality where that happens - and definitely regions of the world where it will - but everywhere..? Eh...

My money for the cynical take is much more along the lines of "massive social upheaval kicks off a world war with terrifying consequences" or "AI eradicates humanity". Less than that and honestly - we're gonna be golden in the long run, and it's gonna be an awkward but neat short/medium run.

Wholeheartedly agreed there. If (still an if) the automation tools are widespread (preferably open source), the vector of possible businesses that could compete in those areas also greatly increases. What once took a firm of thousands of people might be doable by a few nerds with some AI tools they copied from the internet. Yes, that means a lot of jobs lost at those firms, but it also means massive competition to drive prices down to their minimums - which benefits the consumer and enables entirely new second-order effects.

Regardless of how this all plays out... there's gonna a whole lot of Abundance of Things, and Services. As producers/workers - it's gonna be an awkward time. As Consumers? Ho boy baby - shit's gonna get crazy cheap

This is much closer to reality. Markets are priced by demand far more than supply - if you have a perpetual source of demand to milk, the pricing isn't going to come down easily, even if supply could have been made more efficient. I suspect it's a bit of a mixed bag - regulations certainly can be (and are) used as a tool to lock in control in ways that benefit the corporations leading the industry and not the consumer - but abolishing said regulations just makes it even worse. Regardless, if AI is able to lower the bottom-line engineering/labor costs of producing these things (which it absolutely will) then you'll see pretty clearly where the corruption rests. We should expect far better as consumers going forward, and demand both better prices and quality.

That's kind of a good sign of the chokehold weakening though - that desperation only drives more people to the alternatives

Agreed. And fact-checking layers will be coming soon and ya'll don't even want to know what kind of disruption those combined will bring

Eh, doable but you do need the JS layer for updating/changing information dynamically (unless you avoid that sort of feature altogether, or just dump ALL your data onto the page in XML). I'm not that opposed though, really. CSS stuff might get kinda ugly quick as you're essentially doing OOP with it though hmm. This all + a basic JS request/response layer to just fetch more XML (rewriting/overlaying previous data) and you're possibly in business.

Pearl clutchers complaining about the bot that can do pretty much anything with only mediocre quality (so far)

(https://news.ycombinator.com/item?id=33846104)

Agreed. We should also consider the context that - yes - the damages of climate change are disproportionately going to affect developing nations and increase natural disaster risk and food security, but these dangers are coming in against a strong trend of those risks going down for many people. Things were bad in the past - and they're still bad in many places now - but that has been shrinking rapidly. Climate change could very-well slow or even reverse this trend for a time, but barring a complete destruction of our technological progress, people in the future in general will still be far better off than now, even with this mess, and even in developing countries.

Now, of course, this doesn't mean these benefits will be evenly distributed (this is going to even further increase wealth inequality and be a deadly disaster for many people), and we are pretty terrible at prioritizing e.g. protecting natural ecosystem biodiversity. There will certainly be irreversible (in human timescale) losses to our ecosystem and therefore the knowledge and potential of future civilizations from this. And there are even some truly catastrophic death spiral scenarios possible. But it's likely that even the cynical reality will end up being something of a widespread unfairly distributed uptick in disaster risk and continuation of the mass extinction we've already started - slowing quality of life improvements for much of the world, but not stopping them - with an eventual strong recovery in all areas if/when we engage our new levels of technological advancement and begin re-wilding programs to purposefully regrow thriving natural habitats. (This is somewhat inevitable if/when the average quality of life of people is high enough they have room to care about such things. Industrial processes applied to helping nature thrive would do a lot)

This is the "human nature sucks at scale and we're slow to change course" cynical view but with a "technology and industry is crazy good in the long term at achieving its goals and uplifting people" bullish caveat prediction - which I believe to be the age-old trend. We clearly need better global-level response organization to deter the worst effects of climate change and help distribute risk so it's not all falling on the poorest people, but industry-wise - if we have to dump a trillion into something stupid-inefficient like direct carbon capture machines, we will. (My preferred solution is mass-farming kelp - far cheaper per CO2 ton and numerous additional beneficial products that basically make it pay for itself at scale). The world isn't coming to an end. It's coming to a recession/depression - which will be felt to different degrees very unfairly - and it's spurring new responses.

*Latter

Approval voting or proportional representation would be better, but sure