Does seem to be happening more often, recently. E.g. "We need to remind people that they work for the employer, not the other way around" (Tim Gurner) and "People who worked remotely during the pandemic didn’t work as hard – regardless of what th ey told you.” (Stephen Schwarzman)
Sure would be a shame if everyone at Gurner Group decided not to show up for work on the same day. Maybe then Tim Gurner would realize who actually keeps the wheels turning.
But more likely it would be a classic "no...it's the children who are wrong" moment.
I don't understand why people always assume upside and downside risk are symmetrical on the big issues. Consider the set of potential outcomes on a few political hot-buttons:
Global warming (incorrect): We wasted some fraction of human time and resources implementing carbon capture that wasn't necessary. Minor economic hit compared to the counterfactual situation where global warming was never considered.
Global warming (correct): We prevented the collapse of the planet's climate and the ensuing death of most living creatures.
AI will take our jobs (incorrect): We wasted time unnecessarily preparing for a paradigm shift in employment that never happened. Universal basic income was implemented which led to a small positive/negative change in worldwide economic productivity.
AI will take our jobs (correct): UBI was never implemented. A feedback loop emerged where almost all wealth siphoned quickly to a few individuals. The reward mechanism that incentivized the general population to work broke, and hasty attempts by the government to implement a "fair" system of resource distribution among the now "economically worthless" population failed and led to mass supply chain failure, worldwide mental instability, and societal collapse.
Even if you consider the probability of the latter scenarios to be something like 1%, I would argue they are still worth wasting human time and resources on.
> Global warming (incorrect): We wasted some fraction of human time and resources implementing carbon capture that wasn't necessary. Minor economic hit
No, it's not at all a "minor economic" hit with the measures that are being proposed. It is a huge reduction in standard of living for everybody in first world countries, and forcing those who currently don't enjoy a first world standard of living to stay in poverty.
Of course, you could avoid this by building nuclear power plants instead of the draconian carbon reduction/elimination measures that are being proposed. Then we could have the best of both worlds. But no climate change alarmists are proposing that. Why not?
> Global warming (correct): We prevented the collapse of the planet's climate and the ensuing death of most living creatures.
No actual scientific predictions make this claim. It is hype and FUD.
> AI will take our jobs (incorrect): We wasted time unnecessarily preparing for a paradigm shift in employment that never happened. Universal basic income was implemented which led to a small positive/negative change in worldwide economic productivity.
If you think UBI will just be a "small" change, I have some oceanfront property in Tibet I'd like to sell you.
> AI will take our jobs (correct): UBI was never implemented. A feedback loop emerged where almost all wealth siphoned quickly to a few individuals. The reward mechanism that incentivized the general population to work broke, and hasty attempts by the government to implement a "fair" system of resource distribution among the now "economically worthless" population failed and led to mass supply chain failure, worldwide mental instability, and societal collapse.
The consequences you describe here are actually consequences of UBI, not "AI".
The potential negative consequences of "AI" are negative consequences of particular humans using AI as a tool to prey on everyone else--just as such humans have always used whatever tools they can to prey on everyone else. The solution to the problem is not to ban the tool (even if that were possible--you can't ban knowledge). It's to stop letting a few humans prey on everyone else. And the best way to do that is to decentralize power and stop thinking that governments can solve problems. Governments are actually an even better tool than "AI" for some humans to use to prey on everyone else.
> stop thinking that governments can solve problems
In a functional democracy, with rule of law and strong civic institutions, governments is exactly how we solve problems. The more effective the government, the less problems. Compare Europe with the US on policing, food, access to justice, tax filing, banking, fraud prevention, environmental protections, life expectancy, obesity, systems of measurement, public transport, democratic accountability, protection of the elderly, feeding children (I could go on). The US still wrestles with problems that Europe (and much of the rest of the world) has forgotten about - and it does that because it confuses 'freedom' with lack of effective central government.
> In a functional democracy, with rule of law and strong civic institutions, governments is exactly how we solve problems.
Some problems, perhaps. But not all problems. In the US, the government is supposed to be restricted to a certain set of functions that are explicitly delegated to it in the Constitution. It is not supposed to solve everything that anyone thinks is a problem. Of course, the actual practice in the US has diverged greatly from this, but that is a bug, not a feature.
> Compare Europe with the US
I have no problem with Europe having different policies on this from the US. If Europe wants to have government solve many more problems, that's up to Europe. And people who want to live under that model of government should live in Europe.
The US was supposed to be built on a different model of government, for people who want to live under a different model of government from Europe.
> The US still wrestles with problems that Europe (and much of the rest of the world) has forgotten about - and it does that because it confuses 'freedom' with lack of effective central government.
No, the US wrestles with these problems because we prefer to wrestle with them instead of sweeping them under the rug and hoping that "government" will somehow solve them without we the people having to get involved.
I don't share your rosy view of how well Europe has actually solved the problems, but I don't live in Europe so I am not personally affected. And conversely, those who don't live in the US should not presume to tell me how well the US is doing. I live here and I can judge that for myself.
> No, the US wrestles with these problems because we prefer to wrestle with them instead of sweeping them under the rug and hoping that "government" will somehow solve them without we the people having to get involved.
There's this fascinating concept called res publica that means that the public (you, the people) are supposed to do this "government" thing. I've been told you've actually got some of it in the US. Of course, it's entirely possible that what you have is so dysfunctional that the demos have entirely given up their right and responsibility in running the place, but I don't think that's actually true.
While you'll always have a special place in my heart for beta testing the modern version, some of you haven't let that develop into a chauvinism that prevents one from seeing good things elsewhere that might be applicable at home (or for that matter, good things at home which might be applicable elsewhere, because home is completely and utterly incomparable in every way, and one needs not ever even bother finding out) and that's certainly better governance than trying the same thing over and over again, and somehow expecting better results.
I didn't realize how many problems could even exist before I went to the US! A trivial example: Forgetting your card in the ATM - in Europe you have to take your card before you get your cash, so you can't forget it. A very simple UX tweak that solves the problem, that is mandated across all ATMs. Yes a trivial problem, but emblematic of the two different approaches.
The word "supposed" is doing a lot of work in that post.
You can argue anything with the word "supposed" I guess? The voters are "supposed" to vote for elected officials who support policies and programs that will improve their lives? People are "supposed" to find rolemodels who aren't ancient slaveholders? The constitution is "supposed" to be a living document? The federal government is "supposed" to provide for the common good?
The point was about the asymmetry of upside and downside risk. Feel free to insert your own political talking points in place of UBI or carbon capture if you don’t like the specific examples provided for illustrative purposes.
> The point was about the asymmetry of upside and downside risk.
Yes, and I am disputing that point, not just with respect to climate change or other "political talking points", but with respect to AI, which is the topic of the article under discussion.
> But no climate change alarmists are proposing that. Why not?
There are plenty of "climate change alarmists" proposing that. Historically both "green" activists and the corporate profit motive have prevented nuclear deployment (and killed extant nuclear capacity).
> stop thinking that governments can solve problems
Idk, it wasn't corporations who voluntarily chose to stop dumping chemicals into rivers or lead into gasoline. Capitalism is probably the most effective tool to solve 99.999% of problems, but it is not a good tool for the remaining 0.001%, many of which happen to be structurally important, and it's not good for managing its own externalities.
> There are plenty of "climate change alarmists" proposing that.
References, please?
> it wasn't corporations who voluntarily chose to stop dumping chemicals into rivers or lead into gasoline.
It was governments who allowed them to do those things in the first place. The fact that governments eventually stepped in to stop them doesn't mean governments solve problems. It means governments create problems--and then sometimes manage to sort of mitigate them.
> It was governments who allowed them to do those things in the first place. The fact that governments eventually stepped in to stop them doesn't mean governments solve problems. It means governments create problems--and then sometimes manage to sort of mitigate them.
bro what. That's some incredible gymnastics. "Government regulation doesn't solve problems. Examples of problems that government regulation solved are actually examples of the failure of government's prior regulatory policy of 'no regulation,' another damning indictment of regulatory policy!" Remarkable.
> the failure of government's prior regulatory policy of 'no regulation
Bro what?
The government's prior policy was not "no regulation". It was "regulation that prevents ordinary people from challenging corporations because hey, the government's got this, so ordinary people shouldn't be able to stop corporations from building plants that will dump chemicals into rivers or putting toxic substances into gasoline". Which meant those things could be done unchecked until the government decided to stop them--instead of ordinary people getting a chance to say "no" up front, when it would have mattered.
1) What regulation was that? There were all sorts of lawsuits throughout history against polluting companies, most of which were unsuccessful because the companies weren't breaking the law, not because they didn't have a right to sue.
2) Obviously even today, companies can only be sued for... breaking the law. So it's clearly not true that "governments don't solve problems." They create the very structure by which companies can be held accountable for this stuff.
3) What do you mean "when it mattered?" Are you of the belief that it's good ol' self restraint that's stopping companies from dumping chemicals in the rivers today?
You yourself described it: the government-made laws allowed companies to pollute, and that prevented people from successfully suing them--because they weren't breaking the law. Because the government made the laws that way.
> companies can only be sued for... breaking the law.
In our current government-controlled environment, yes, this is true. But that is no argument for our current government-controlled environment being a good thing.
> What do you mean "when it mattered?"
Isn't that obvious? When it mattered was before the polluting started in the first place. What prevents people from stopping these things in advance? The fact that they don't own the land and the rivers and the air that are being polluted--the government does. So only the government can stop them from being polluted. The government has taken away the people's ownership of these things and is doing a very bad job of stewardship of them. And you think this is good?
The singularity scenario does not require malice to result in disaster. When your capabilities are superhuman, almost every goal is more easily and reliably accomplished with no living creatures around to complicate things. Unless the goal explicitly preserves them, the logical option is to kill them all. And "accomplish goal", or "accomplish goal with a botched constraint that doesn't actually save anyone" is simpler than "accomplish goal without killing everybody", so it's more likely to be tried first.
With global warming - a clear and present danger, the only realistic immediate and effective solution is to stop digging up fossil fuels and burning them. However almost all the world's economy is predicated on precisely that, so it is a big ask.
With AI - a nebulous threat based on not much more than some 21st century parlor tricks, it is quite reasonable to 'cross that bridge when we reach it' (which may be never).
>it is quite reasonable to 'cross that bridge when we reach it' (which may be never).
Unless you are a junior copy writer... Or a midlevel photographer... Or production artist... You better cross that bridge before your next rent check is due.
The problem with both those is that the possibilities are not simple correct/incorrect sets of predictable outcomes.
There are a whole lot of probable scenarios and possible responses. What is the right response? How do we execute it?
Take global warming. What if we have already passed some tipping point and its too late to do anything? Are current responses adequate? Are we adopting the right technologies to replace fossil fuels? Both your examples of "incorrect" (which I am interpreting as global warming has no bad effects whatsoever) and "death of most living creatures" are extreme and highly improbable.
AI is even worse as we do not know what the possible outcomes are - we do not know what the technology is capable of, or how it will develop.
I agree that UBI will do good either way. However, I do not think your AI scenario is that likely. It will not lead to supply chain failure, for example - the work will still be done. It will also not be a sudden change as "AI" is not going to take over everyone's jobs suddenly - it will be a gradual process.
> What if we have already passed some tipping point and its too late to do anything?
Even if we've already passed a number of tipping points, the choice between "worst case" and "less so" is the difference between millions, even billions of lives.
Somewhat off-topic, but I must ask: does UBI not depend on expected infinite economic growth?
If the economy grows, then we assume the tax base supplanting the UBI coffers will continue to match outflows. But if the economy does not grow, then... what happens? The starkest issue in my mind is what is happening in the EU and the US: social welfare is not looking sustainable from an economic perspective.
And if we do implement UBI to such a degree that people are allowed to live a life completely without any physical need for traditional income... wouldn't that cause the collapse of the country's economy when its workers realize there is no longer a point to laboring? At that point, who exactly is going to pay for UBI?
I don't believe I have forgotten. I'm assuming it will be able to pay for not-great-but-liveable housing, sufficient food, and healthcare. That is a life that I consider to be completely without any physical need for traditional income.
Granted, there are people that want more material things, and they would be motivated to labor for those things. However, my intuition tells me they're not a majority -- or enough to continue supplanting the UBI coffers.
I feel like a majority of people want to be able to afford new clothes, entertainment, and occasional luxuries like a dinner out or travel. Even if we have people who labor intermittently to save up for their next purchase like that you'd still see a pool of labor from that.
One quandary I have that might be interesting to explore is: are the material requirements of modern day people living in Western civilization something that is innate to humans' propensity for wanting more or something that has come about due to a variety of environmental factors that drove us to this outcome. And how will that shift if and when UBI is implemented?
But I think I've replied too many times already; so I'll let someone else have at it, and stop my chattering.
What do you consider "sufficient"? If it's paying you just enough so you can have milk and eggs and bread every day, is that sufficient? Do you really think people would be willing to live a monotonous life for the rest of their lives?
The point of UBI isn't to make it rain so that nobody has to work anymore. It's to make it so that nobody ends up on the streets just because they lost their job for a while. And a pleasant side effect is people will be able to invest more time into finding better jobs, educating themselves in new areas, etc. without fear of going homeless or broke.
I'm not claiming funding isn't an issue (it most definitely is, as are a bunch of other things), but you're mixing up the funding aspect with the motivation aspect, which need not be tied so strongly together.
I think one possibility is that the increase in technology is enough to cover our needs.
For example forget money, but instead we just handed out a certain amount of food to everybody. We can get 1 person with an advance combine to collect enough grain for maybe 1000 people. If we can automate that, then 0 people collect enough grain for infinite people. The only limit is the area that we are farming.
It would be reasonable to assume for ubi to work, we would need a reasonable amount of land to be farmed.
Now apply the same concept to other things as well.
That's not really a step up from the current system that lifted millions out of poverty and hunger.
The US has enough grain to pass it out to everyone, no magic technological leap needed, we already ship our massive amounts of excess grain across the world to developing countries.
Noone in a developed country wants to eat just grain though!
Also who gets to hand out the grain? That never really goes well when authoritarians control who gets to eat and how much and what. (some animals are more equal than others)
And who gets / has to do the automating / maintaining? Are they compensated extra or is it forced?
> It would be reasonable to assume for ubi to work, we would need a reasonable amount of land to be farmed.
We have plenty of farmland. The thing you need most for "free food" is more ranchland in the US.
Otherwise you will have to develop the natural grassland as farmland, killing every creature under the topsoil and using A LOT of water.
As you can tell I have a few issues with this communistic approach, but I would like honest answers to these proses as they come up in history as "problems of implementation" to be generous. (killing millions via famine and force)
Edit: downvote all you want, but questioning a proposed new system to govern humans should be scrutinized given the past failed experiments with millions of people's lives.
This is an interesting thought. And at that point, if we have sufficiently advanced technology we can simply do away with money and dispense necessities as necessary. A bit like communism, if it hadn't been plagued by resource shortages (ignoring the human causes).
>> If we can automate that, then 0 people collect enough grain for infinite people.
The people who built the automation want to be paid (in perpetuity since they "own" it), as do the people who maintain the automation. But when nobody has a job, they can't pay. So we confiscate from the "owner" and tax the wages of the "maintainers" and hand it to the masses via UBI so they can "pay" for the services of the others (but food is free). I suspect wealth redistribution is an important function of the US government and tax system, but when things get to far out of whack it's going to look absurdly stupid and people will not play any more.
You absolutely do not need exponential growth for UBI in a world with heavy automation. UBI is needed in a world where a large percentage of jobs have been automated, that implies everything in your supply chain is run by robots.
When robots produce the goods, UBI is just a mechanism to distribute those goods evenly among society.
> Somewhat off-topic, but I must ask: does UBI not depend on expected infinite economic growth?
No, a Universal Basic Income need not depend on infinite economic growth, and in its simplest form, it is not guaranteed to be constant, only universal. Variable subsidies are no more difficult to accommodate than variable taxes when they are similarly grounded in the economy, and while economic growth would certainly contribute to the funding of UBI, stable revenue streams exist that could, theoretically, maintain its sustainability in perpetuity. In its original conception as the Citizen's Dividend[1], UBI was to be funded in its entirety by taxing price appreciation on rivalrous inelastic goods, or "land" in the economic sense[2], which includes things like electromagnetic spectrum, clean air, seignorage, and yes, physical land. (This is the infamous "Land Value Tax" that has been making the rounds lately[3].) As such, properly implemented, UBI can be entirely nondistortionary.
> [...] wouldn't that cause the collapse of the country's economy when its workers realize there is no longer a point to laboring? At that point, who exactly is going to pay for UBI?
The way the finances work out, by the time taxable economic rents are large enough to support most of Maslow's hierarchy via UBI for all of Earth's citizens, it won't matter. At that point, energy is likely abundant enough so as to be free, and the intergalactic market economy would be little more than a game we play to keep ourselves entertained. In other words, this is a problem so far down the road that we're guaranteed to have far bigger problems to worry about in the meantime, and elevating the standard of living for as many people as possible would only help in dealing with them. Also, keep in mind, few are happy with merely surviving, and there exist many definitions of thriving. These inherent disparities in human desire are more than enough to keep the game going.
ubi just shifts the income distribution so a larger fraction gets spent on the things the poor want than used to be the case. e.g. Fewer McMansions, more apartments. the incentive to work isn't suddenly gone, just no longer life threatening.
compare to the standard conservative argument for inequality (a rising tide lifts all boats), which explicitly depends on economic growth.
> Global warming (incorrect): We wasted some fraction of human time and resources implementing carbon capture that wasn't necessary. Minor economic hit compared to the counterfactual situation where global warming was never considered.
Consider carefully: what would happen if the massive body of corroborating evidence for global warming-- across many disparate fields of scientific inquiry-- was incorrect? We'd likely get a massive rejection of the sciences as a whole that makes current anti-science trends look minor by comparison.
E.g., notice how many low-effort HN commenters currently reject social science out of hand because of the risk of P hacking. Imagine what happens when that goes into geology, biology, etc.
Also, consider that Pentagon studies that make important predictions about refugee movements based partly on the same science.
An evidence-based rejection of so many key institutions of civilization wouldn't be as catastrophic as doing nothing if global warming is correct, but it's a close second.
Maybe I'm missing a steelman argument here? It's really difficult to imagine a version of "Global warming (incorrect)" which doesn't substantially depend on our knowledge of global warming being correct and doing something about it.
Edit: Just be clear, I feel like this Pascal's wager argument is a bit like trying to talk a pyramid schemer down by explaining exponential explosion to them. You've got to realize at that point if it were simply about logic you probably wouldn't have an adversary to argue with at all.
The Spaghetti Monster doesn't exist: we wasted some fraction of human time and resources worshipping the Spaghetti monster.
The Spaghetti Monster exists: we gain access to the Pastafarian Heaven for eternity.
Surely we should all convert? Even if the probability of the Spaghetti Monster existing is very low it's still worth wasting just a little bit of time and resources on prayers.
The other side of the wager is rarely considered (AKA Homer’s Wager):
What if the Udon Noodle Monster exist and he intensely hates people who worship someone else instead of him but is neutral on people who don’t worship at all.
Simple solution: consider the universal a priori probability of monster X as the negative exponentially weighted size of the minimum length Turing machine necessary to describe the monster, normalized against all other possible monsters Y, Z, etc. (if the monster isn’t describable by a computable process, we can’t change our behavior with regard to it so we’re screwed either way.)
Then, consider a “reward program” that takes as input the computable description of a monster and returns the afterlife reward as a function of worship time. (There is some nuance here; should reward be defined as negative aggregate suffering? Median pleasure over afterlife duration? Do we include other people’s suffering in exchange for our own? Different debate.)
At this point, you consider all possible divisions of finite worship time among the infinite set of monsters whose descriptions are partial recursive functions and compute the probability distribution of reward for each configuration using the universal a priori probabilities of monster existence (while taking into account second order effects, e.g. monster A reduces reward for time spent worshiping monster B). From here, you specify a minimum threshold of acceptable reward and select the configuration of worship time that corresponds to the highest probability of achieving that threshold.
A person comes up to you and tells you "I am a god. Give me your wallet or you will face eternal suffering in the afterlife."
The person isnt a god: you lose the $50 you had in your wallet and an afternoon requesting new cards / IDs if you accept, and nothing happens if you refuse
The person is a god: you lose the time and money if you accept, and face the worst possible punishment if you refuse
Surely the only logical option is to give the person your wallet!
(Note: I do agree with the original OP of this comment chain that it is worth the cost of climate mitigation. Climate change mitigation is not a kind of Pascal's Wager because global warming is scientific fact. We know with certainty that it will continue to happen without a course change, and that this will cause problems for humanity and the environment)
Pascal's wager doesn't sound too bad for the 1% chance I mentioned. 0.0001% chance? Maybe not so much.
Now I suppose you could argue there are an infinite number of scenarios where there is a 1% chance of some kind of disproportionate reward, or conversely, a 1% chance of an outsized penalty. Expected value isn't really the right target to optimize for anyway, otherwise we might all be buying lottery tickets when the jackpot becomes large enough.
If I have to formalize my post a bit more, I'd say we first narrow down the set of issues into those where the population is roughly split on the matter. Even if you have a very compelling case for why an asteroid is going to strike the earth in the next year, if you can't convince enough people to adopt your viewpoint then it's not going to make any difference, and you may as well optimize for enjoying your last year alive instead.
Then once we have this set of critical issues, we allow each side to consider the probability and associated damage of every outcome resulting from the set of decisions to be made about each issue. Why do both sides of an issue get equal representation? Shouldn't the more scientifically literate side carry more weight? Again, the problem here is not devising the best predictive model of outcomes but rather convincing a majority to buy into a particular decision making framework that allows for a significant reduction in overall societal risk.
Once each side has provided their subjective confidence of the various outcomes and the resulting damage, then you specify a threshold for acceptable damage. For example, what amount of damage to the economy would be expected due to attempts to mitigate global warming? What amount of damage to the climate would be expected if nothing were done? I would think most people's baseline threshold for acceptable risk is similar even if they disagree on the particular issues. At this point, each side should then be satisfied with the knowledge that even though the other side is "wrong", they got their way on some things and the other side got their way on some things in a manner that reduces each side's subjective global risk to a tolerable amount.
The 'incorrect' version of global warming is that we fall behind other countries in all measures because we're focusing on a non-issue. One of those invades us and takes us over.
“And anyway most people like to be employed” is exactly the kind of room temperature iq take I would expect from someone who is equating ai putting people out of work with the rise of the equine glue industry without seeming to have any appreciation for the horrific reality they are alluding to. It’s amazing to me that people who are supposed to have foresight to navigate their companies have so little imagination.
As a point of housekeeping, the title here was a little confusing to read. A lot of negatives. Reading the article seems to give a completely different story as a telco exec likens the horse-to-car transition as similar to what we're experiencing now with AI/ Not sure how the title fits in to that.
"Just because you're paranoid, doesn't mean they're not out to get you" is a standard saying (with "they" left undefined, and fitting in nicely with the way paranoids often talk). The headline is riffing on that saying.
I consider AI as still at the early stage, if we're talking about chees AI that began it's popularity in 80s and the surge of LLM at 2018. Yet just after several years since chatGPT, I see that there's many drawings of AI-generated that's good. The growth of AI is astonishing fast and in 5 years it'll be completely different beast than today.
I think people underestimate the capability of AI. It's effect may be equal with or more than industrial revolution. Mass surveillance, deep fakes, text to voices, ocr, autonomous driving, will be conducted at a scale that never be seen until now. This may be the first time where a small set of party can hold physical (like in military) powers over millions of people, after nukes.
If the promises behind the current hype on AI are correct, this means the end of the middle classes, the end of any chance of social mobility. So, it is going to accelerate things that were already in motion, but by a massive factor.
And it is pretty ironic that a lot of the capital that went into this was the result of massive monetary expansion that led initially to massive asset inflation and later to garden variety inflation. The thing with inflation is that is never distributively neutral, it is always a wealth transfer engine.
Richer people in a inflationary expansion are able to capture a lot of the early money, and so the capital holding class are able to do things like dabbling first on self-driving cars, and now on massive LL Models.
So, basically, we are transfering income and wealth from the poor to enable the rich to get rid of their need of buying labor from the poor.
How this doesn't end in a dystopian dictatorship beyond any of the worst nightmares of people like Orwell, or else in a Frank Hebert's "Butlerian Jihad" is something that I am at a loss of imagination to see.
A better analogy IMO is the transition from subsistence to industrial agriculture. Once upon a time, at least half the U.S. population worked in agricultural fields - then, to condense a lot of history, tractors and seed implanters and combine harvesters were invented, and now something like 2% of the U.S. population is directly involved in growing food (although certain vegetable and fruit crops are much more labor-intensive).
I've never heard anybody bemoaning this development - agricultural field jobs are back-breaking and dull - but it did create a situation of greater societal dependence on a much smaller number of farmers. Ensuring that everyone gets enough food then becomes something of a socioeconomic problem rather than a technical problem.
Now, if LLM-related technology starts making major inroads into the well-paid upper-tier employment sector, this could get a bit more interesting. My favorite suggestion is to use LLMs to replace shareholder-appointed corporate board members - and hey, why not the shareholders themselves? LLMs might be much better at generating capital from investments, and as they're not wasting the money on giant yachts and private islands, they might be much more efficient at plowing capital gains back into research and development and so improving manufacturing.
This sort of highlights the fact that a society in which all the wealth is in the hands of a few aristocrats is a society doomed to eventual extinction, particularly in an era of rapid technological innovation.
I often feel like some people around me have been programmed by the AI as they're constantly regurgitating some trendy mainstream narratives word-for-word. I increasingly feel like an 'Outlier' from the series Westworld.
I think it's probably because we're constantly bombarded with mainstream narratives and people in our peer group are shown the same news articles and content at the same time. People who actually believe the articles/mainstream narratives tend to come across (to the rest of us) as though they've been programmed because they're all saying and thinking about the same thing at the same time in perfect unison. There's something eery about that.
AI and capitalism are an incredibly dangerous combination.
Associating your human worth with your economic productivity and then limiting exponential economic productivity to a few already powerful people is a nightmare.
Economies of scale can be rephrased as winner take all.
Never thought I'd say this, but I am glad post modernism exists. A philosophical movement that rejects grounded representations of art or philosophy is somehow the perfect bulwark against AI.
Why would it be limited to a few people? If the impacts are as far reaching as people expect an open source AGI will be inevitable. Useful algorithms are very hard to keep secret.
I imagine a future with few jobs, where everyone is largely self sufficient. Would that really be so bad?
Doesn't matter if AGI is open source. The main problem is the devaluing of human labor which a lot of people today trade for vital goods and services.
People who own nothing that can produce continuous value will not have anything to trade with in the long term and will effectively be excluded from the economy.
You're describing a post scarcity world. The whole point of an economy is to manage scarcity, talking about economic value post AGI is nonsensical. The responsibility to produce and distribute goods post AGI will fall to the government. Your doomsday scenario only exists in a world where the government purposefully chooses to deprive its citizens.
People who don't want to depend on the government could always spin up their own AGI and become self sufficient. Some solar panels, a few robots and some farmable land are probably all you need when it comes down to it.
It’s not post-scarcity. There are still limits to the amount of material goods we can produce even though services are effectively free - but not completely free.
People who own the means of production, the factories, the mines, the farms, the power stations, … etc. will control who they sell their goods to and for how much.
Having to rely on the government is scary. If you got a bad government you are so screwed. You have nigh zero leverage as you have nothing to offer anyone.
76 comments
[ 3.5 ms ] story [ 193 ms ] threadIt's not that often a member of the executive class forgets themselves and says the quiet part this loudly: to them, we are livestock.
But more likely it would be a classic "no...it's the children who are wrong" moment.
Global warming (incorrect): We wasted some fraction of human time and resources implementing carbon capture that wasn't necessary. Minor economic hit compared to the counterfactual situation where global warming was never considered.
Global warming (correct): We prevented the collapse of the planet's climate and the ensuing death of most living creatures.
AI will take our jobs (incorrect): We wasted time unnecessarily preparing for a paradigm shift in employment that never happened. Universal basic income was implemented which led to a small positive/negative change in worldwide economic productivity.
AI will take our jobs (correct): UBI was never implemented. A feedback loop emerged where almost all wealth siphoned quickly to a few individuals. The reward mechanism that incentivized the general population to work broke, and hasty attempts by the government to implement a "fair" system of resource distribution among the now "economically worthless" population failed and led to mass supply chain failure, worldwide mental instability, and societal collapse.
Even if you consider the probability of the latter scenarios to be something like 1%, I would argue they are still worth wasting human time and resources on.
No, it's not at all a "minor economic" hit with the measures that are being proposed. It is a huge reduction in standard of living for everybody in first world countries, and forcing those who currently don't enjoy a first world standard of living to stay in poverty.
Of course, you could avoid this by building nuclear power plants instead of the draconian carbon reduction/elimination measures that are being proposed. Then we could have the best of both worlds. But no climate change alarmists are proposing that. Why not?
> Global warming (correct): We prevented the collapse of the planet's climate and the ensuing death of most living creatures.
No actual scientific predictions make this claim. It is hype and FUD.
> AI will take our jobs (incorrect): We wasted time unnecessarily preparing for a paradigm shift in employment that never happened. Universal basic income was implemented which led to a small positive/negative change in worldwide economic productivity.
If you think UBI will just be a "small" change, I have some oceanfront property in Tibet I'd like to sell you.
> AI will take our jobs (correct): UBI was never implemented. A feedback loop emerged where almost all wealth siphoned quickly to a few individuals. The reward mechanism that incentivized the general population to work broke, and hasty attempts by the government to implement a "fair" system of resource distribution among the now "economically worthless" population failed and led to mass supply chain failure, worldwide mental instability, and societal collapse.
The consequences you describe here are actually consequences of UBI, not "AI".
The potential negative consequences of "AI" are negative consequences of particular humans using AI as a tool to prey on everyone else--just as such humans have always used whatever tools they can to prey on everyone else. The solution to the problem is not to ban the tool (even if that were possible--you can't ban knowledge). It's to stop letting a few humans prey on everyone else. And the best way to do that is to decentralize power and stop thinking that governments can solve problems. Governments are actually an even better tool than "AI" for some humans to use to prey on everyone else.
In a functional democracy, with rule of law and strong civic institutions, governments is exactly how we solve problems. The more effective the government, the less problems. Compare Europe with the US on policing, food, access to justice, tax filing, banking, fraud prevention, environmental protections, life expectancy, obesity, systems of measurement, public transport, democratic accountability, protection of the elderly, feeding children (I could go on). The US still wrestles with problems that Europe (and much of the rest of the world) has forgotten about - and it does that because it confuses 'freedom' with lack of effective central government.
Some problems, perhaps. But not all problems. In the US, the government is supposed to be restricted to a certain set of functions that are explicitly delegated to it in the Constitution. It is not supposed to solve everything that anyone thinks is a problem. Of course, the actual practice in the US has diverged greatly from this, but that is a bug, not a feature.
> Compare Europe with the US
I have no problem with Europe having different policies on this from the US. If Europe wants to have government solve many more problems, that's up to Europe. And people who want to live under that model of government should live in Europe.
The US was supposed to be built on a different model of government, for people who want to live under a different model of government from Europe.
> The US still wrestles with problems that Europe (and much of the rest of the world) has forgotten about - and it does that because it confuses 'freedom' with lack of effective central government.
No, the US wrestles with these problems because we prefer to wrestle with them instead of sweeping them under the rug and hoping that "government" will somehow solve them without we the people having to get involved.
I don't share your rosy view of how well Europe has actually solved the problems, but I don't live in Europe so I am not personally affected. And conversely, those who don't live in the US should not presume to tell me how well the US is doing. I live here and I can judge that for myself.
There's this fascinating concept called res publica that means that the public (you, the people) are supposed to do this "government" thing. I've been told you've actually got some of it in the US. Of course, it's entirely possible that what you have is so dysfunctional that the demos have entirely given up their right and responsibility in running the place, but I don't think that's actually true.
While you'll always have a special place in my heart for beta testing the modern version, some of you haven't let that develop into a chauvinism that prevents one from seeing good things elsewhere that might be applicable at home (or for that matter, good things at home which might be applicable elsewhere, because home is completely and utterly incomparable in every way, and one needs not ever even bother finding out) and that's certainly better governance than trying the same thing over and over again, and somehow expecting better results.
I didn't realize how many problems could even exist before I went to the US! A trivial example: Forgetting your card in the ATM - in Europe you have to take your card before you get your cash, so you can't forget it. A very simple UX tweak that solves the problem, that is mandated across all ATMs. Yes a trivial problem, but emblematic of the two different approaches.
You can argue anything with the word "supposed" I guess? The voters are "supposed" to vote for elected officials who support policies and programs that will improve their lives? People are "supposed" to find rolemodels who aren't ancient slaveholders? The constitution is "supposed" to be a living document? The federal government is "supposed" to provide for the common good?
Yes, and I am disputing that point, not just with respect to climate change or other "political talking points", but with respect to AI, which is the topic of the article under discussion.
There are plenty of "climate change alarmists" proposing that. Historically both "green" activists and the corporate profit motive have prevented nuclear deployment (and killed extant nuclear capacity).
> stop thinking that governments can solve problems
Idk, it wasn't corporations who voluntarily chose to stop dumping chemicals into rivers or lead into gasoline. Capitalism is probably the most effective tool to solve 99.999% of problems, but it is not a good tool for the remaining 0.001%, many of which happen to be structurally important, and it's not good for managing its own externalities.
References, please?
> it wasn't corporations who voluntarily chose to stop dumping chemicals into rivers or lead into gasoline.
It was governments who allowed them to do those things in the first place. The fact that governments eventually stepped in to stop them doesn't mean governments solve problems. It means governments create problems--and then sometimes manage to sort of mitigate them.
> It was governments who allowed them to do those things in the first place. The fact that governments eventually stepped in to stop them doesn't mean governments solve problems. It means governments create problems--and then sometimes manage to sort of mitigate them.
bro what. That's some incredible gymnastics. "Government regulation doesn't solve problems. Examples of problems that government regulation solved are actually examples of the failure of government's prior regulatory policy of 'no regulation,' another damning indictment of regulatory policy!" Remarkable.
Bro what?
The government's prior policy was not "no regulation". It was "regulation that prevents ordinary people from challenging corporations because hey, the government's got this, so ordinary people shouldn't be able to stop corporations from building plants that will dump chemicals into rivers or putting toxic substances into gasoline". Which meant those things could be done unchecked until the government decided to stop them--instead of ordinary people getting a chance to say "no" up front, when it would have mattered.
2) Obviously even today, companies can only be sued for... breaking the law. So it's clearly not true that "governments don't solve problems." They create the very structure by which companies can be held accountable for this stuff.
3) What do you mean "when it mattered?" Are you of the belief that it's good ol' self restraint that's stopping companies from dumping chemicals in the rivers today?
You yourself described it: the government-made laws allowed companies to pollute, and that prevented people from successfully suing them--because they weren't breaking the law. Because the government made the laws that way.
> companies can only be sued for... breaking the law.
In our current government-controlled environment, yes, this is true. But that is no argument for our current government-controlled environment being a good thing.
> What do you mean "when it mattered?"
Isn't that obvious? When it mattered was before the polluting started in the first place. What prevents people from stopping these things in advance? The fact that they don't own the land and the rivers and the air that are being polluted--the government does. So only the government can stop them from being polluted. The government has taken away the people's ownership of these things and is doing a very bad job of stewardship of them. And you think this is good?
With AI - a nebulous threat based on not much more than some 21st century parlor tricks, it is quite reasonable to 'cross that bridge when we reach it' (which may be never).
Unless you are a junior copy writer... Or a midlevel photographer... Or production artist... You better cross that bridge before your next rent check is due.
There are a whole lot of probable scenarios and possible responses. What is the right response? How do we execute it?
Take global warming. What if we have already passed some tipping point and its too late to do anything? Are current responses adequate? Are we adopting the right technologies to replace fossil fuels? Both your examples of "incorrect" (which I am interpreting as global warming has no bad effects whatsoever) and "death of most living creatures" are extreme and highly improbable.
AI is even worse as we do not know what the possible outcomes are - we do not know what the technology is capable of, or how it will develop.
I agree that UBI will do good either way. However, I do not think your AI scenario is that likely. It will not lead to supply chain failure, for example - the work will still be done. It will also not be a sudden change as "AI" is not going to take over everyone's jobs suddenly - it will be a gradual process.
Even if we've already passed a number of tipping points, the choice between "worst case" and "less so" is the difference between millions, even billions of lives.
If the economy grows, then we assume the tax base supplanting the UBI coffers will continue to match outflows. But if the economy does not grow, then... what happens? The starkest issue in my mind is what is happening in the EU and the US: social welfare is not looking sustainable from an economic perspective.
And if we do implement UBI to such a degree that people are allowed to live a life completely without any physical need for traditional income... wouldn't that cause the collapse of the country's economy when its workers realize there is no longer a point to laboring? At that point, who exactly is going to pay for UBI?
I think you're forgetting the B in UBI. The idea is to guarantee basic necessities, not to maintain a high quality of life.
Granted, there are people that want more material things, and they would be motivated to labor for those things. However, my intuition tells me they're not a majority -- or enough to continue supplanting the UBI coffers.
One quandary I have that might be interesting to explore is: are the material requirements of modern day people living in Western civilization something that is innate to humans' propensity for wanting more or something that has come about due to a variety of environmental factors that drove us to this outcome. And how will that shift if and when UBI is implemented?
But I think I've replied too many times already; so I'll let someone else have at it, and stop my chattering.
The point of UBI isn't to make it rain so that nobody has to work anymore. It's to make it so that nobody ends up on the streets just because they lost their job for a while. And a pleasant side effect is people will be able to invest more time into finding better jobs, educating themselves in new areas, etc. without fear of going homeless or broke.
I'm not claiming funding isn't an issue (it most definitely is, as are a bunch of other things), but you're mixing up the funding aspect with the motivation aspect, which need not be tied so strongly together.
It would be reasonable to assume for ubi to work, we would need a reasonable amount of land to be farmed.
Now apply the same concept to other things as well.
That's not really a step up from the current system that lifted millions out of poverty and hunger.
The US has enough grain to pass it out to everyone, no magic technological leap needed, we already ship our massive amounts of excess grain across the world to developing countries.
Noone in a developed country wants to eat just grain though!
Also who gets to hand out the grain? That never really goes well when authoritarians control who gets to eat and how much and what. (some animals are more equal than others)
And who gets / has to do the automating / maintaining? Are they compensated extra or is it forced?
> It would be reasonable to assume for ubi to work, we would need a reasonable amount of land to be farmed.
We have plenty of farmland. The thing you need most for "free food" is more ranchland in the US.
Otherwise you will have to develop the natural grassland as farmland, killing every creature under the topsoil and using A LOT of water.
As you can tell I have a few issues with this communistic approach, but I would like honest answers to these proses as they come up in history as "problems of implementation" to be generous. (killing millions via famine and force)
Edit: downvote all you want, but questioning a proposed new system to govern humans should be scrutinized given the past failed experiments with millions of people's lives.
The people who built the automation want to be paid (in perpetuity since they "own" it), as do the people who maintain the automation. But when nobody has a job, they can't pay. So we confiscate from the "owner" and tax the wages of the "maintainers" and hand it to the masses via UBI so they can "pay" for the services of the others (but food is free). I suspect wealth redistribution is an important function of the US government and tax system, but when things get to far out of whack it's going to look absurdly stupid and people will not play any more.
When robots produce the goods, UBI is just a mechanism to distribute those goods evenly among society.
No, a Universal Basic Income need not depend on infinite economic growth, and in its simplest form, it is not guaranteed to be constant, only universal. Variable subsidies are no more difficult to accommodate than variable taxes when they are similarly grounded in the economy, and while economic growth would certainly contribute to the funding of UBI, stable revenue streams exist that could, theoretically, maintain its sustainability in perpetuity. In its original conception as the Citizen's Dividend[1], UBI was to be funded in its entirety by taxing price appreciation on rivalrous inelastic goods, or "land" in the economic sense[2], which includes things like electromagnetic spectrum, clean air, seignorage, and yes, physical land. (This is the infamous "Land Value Tax" that has been making the rounds lately[3].) As such, properly implemented, UBI can be entirely nondistortionary.
> [...] wouldn't that cause the collapse of the country's economy when its workers realize there is no longer a point to laboring? At that point, who exactly is going to pay for UBI?
The way the finances work out, by the time taxable economic rents are large enough to support most of Maslow's hierarchy via UBI for all of Earth's citizens, it won't matter. At that point, energy is likely abundant enough so as to be free, and the intergalactic market economy would be little more than a game we play to keep ourselves entertained. In other words, this is a problem so far down the road that we're guaranteed to have far bigger problems to worry about in the meantime, and elevating the standard of living for as many people as possible would only help in dealing with them. Also, keep in mind, few are happy with merely surviving, and there exist many definitions of thriving. These inherent disparities in human desire are more than enough to keep the game going.
[1] https://en.wikipedia.org/wiki/Citizen%27s_dividend
[2] https://en.wikipedia.org/wiki/Land_(economics)
[3] https://en.wikipedia.org/wiki/Land_value_tax
compare to the standard conservative argument for inequality (a rising tide lifts all boats), which explicitly depends on economic growth.
Consider carefully: what would happen if the massive body of corroborating evidence for global warming-- across many disparate fields of scientific inquiry-- was incorrect? We'd likely get a massive rejection of the sciences as a whole that makes current anti-science trends look minor by comparison.
E.g., notice how many low-effort HN commenters currently reject social science out of hand because of the risk of P hacking. Imagine what happens when that goes into geology, biology, etc.
Also, consider that Pentagon studies that make important predictions about refugee movements based partly on the same science.
An evidence-based rejection of so many key institutions of civilization wouldn't be as catastrophic as doing nothing if global warming is correct, but it's a close second.
Maybe I'm missing a steelman argument here? It's really difficult to imagine a version of "Global warming (incorrect)" which doesn't substantially depend on our knowledge of global warming being correct and doing something about it.
Edit: Just be clear, I feel like this Pascal's wager argument is a bit like trying to talk a pyramid schemer down by explaining exponential explosion to them. You've got to realize at that point if it were simply about logic you probably wouldn't have an adversary to argue with at all.
The Spaghetti Monster doesn't exist: we wasted some fraction of human time and resources worshipping the Spaghetti monster.
The Spaghetti Monster exists: we gain access to the Pastafarian Heaven for eternity.
Surely we should all convert? Even if the probability of the Spaghetti Monster existing is very low it's still worth wasting just a little bit of time and resources on prayers.
What if the Udon Noodle Monster exist and he intensely hates people who worship someone else instead of him but is neutral on people who don’t worship at all.
Then, consider a “reward program” that takes as input the computable description of a monster and returns the afterlife reward as a function of worship time. (There is some nuance here; should reward be defined as negative aggregate suffering? Median pleasure over afterlife duration? Do we include other people’s suffering in exchange for our own? Different debate.)
At this point, you consider all possible divisions of finite worship time among the infinite set of monsters whose descriptions are partial recursive functions and compute the probability distribution of reward for each configuration using the universal a priori probabilities of monster existence (while taking into account second order effects, e.g. monster A reduces reward for time spent worshiping monster B). From here, you specify a minimum threshold of acceptable reward and select the configuration of worship time that corresponds to the highest probability of achieving that threshold.
A person comes up to you and tells you "I am a god. Give me your wallet or you will face eternal suffering in the afterlife."
The person isnt a god: you lose the $50 you had in your wallet and an afternoon requesting new cards / IDs if you accept, and nothing happens if you refuse
The person is a god: you lose the time and money if you accept, and face the worst possible punishment if you refuse
Surely the only logical option is to give the person your wallet!
(Note: I do agree with the original OP of this comment chain that it is worth the cost of climate mitigation. Climate change mitigation is not a kind of Pascal's Wager because global warming is scientific fact. We know with certainty that it will continue to happen without a course change, and that this will cause problems for humanity and the environment)
Now I suppose you could argue there are an infinite number of scenarios where there is a 1% chance of some kind of disproportionate reward, or conversely, a 1% chance of an outsized penalty. Expected value isn't really the right target to optimize for anyway, otherwise we might all be buying lottery tickets when the jackpot becomes large enough.
If I have to formalize my post a bit more, I'd say we first narrow down the set of issues into those where the population is roughly split on the matter. Even if you have a very compelling case for why an asteroid is going to strike the earth in the next year, if you can't convince enough people to adopt your viewpoint then it's not going to make any difference, and you may as well optimize for enjoying your last year alive instead.
Then once we have this set of critical issues, we allow each side to consider the probability and associated damage of every outcome resulting from the set of decisions to be made about each issue. Why do both sides of an issue get equal representation? Shouldn't the more scientifically literate side carry more weight? Again, the problem here is not devising the best predictive model of outcomes but rather convincing a majority to buy into a particular decision making framework that allows for a significant reduction in overall societal risk.
Once each side has provided their subjective confidence of the various outcomes and the resulting damage, then you specify a threshold for acceptable damage. For example, what amount of damage to the economy would be expected due to attempts to mitigate global warming? What amount of damage to the climate would be expected if nothing were done? I would think most people's baseline threshold for acceptable risk is similar even if they disagree on the particular issues. At this point, each side should then be satisfied with the knowledge that even though the other side is "wrong", they got their way on some things and the other side got their way on some things in a manner that reduces each side's subjective global risk to a tolerable amount.
The 'incorrect' version of global warming is that we fall behind other countries in all measures because we're focusing on a non-issue. One of those invades us and takes us over.
I think people underestimate the capability of AI. It's effect may be equal with or more than industrial revolution. Mass surveillance, deep fakes, text to voices, ocr, autonomous driving, will be conducted at a scale that never be seen until now. This may be the first time where a small set of party can hold physical (like in military) powers over millions of people, after nukes.
Actually it was just one year ago.
Jokes aside, points still stand.
Guaranteed though, more of them were eaten or sent to the glue factory.
And it is pretty ironic that a lot of the capital that went into this was the result of massive monetary expansion that led initially to massive asset inflation and later to garden variety inflation. The thing with inflation is that is never distributively neutral, it is always a wealth transfer engine.
Richer people in a inflationary expansion are able to capture a lot of the early money, and so the capital holding class are able to do things like dabbling first on self-driving cars, and now on massive LL Models.
So, basically, we are transfering income and wealth from the poor to enable the rich to get rid of their need of buying labor from the poor.
How this doesn't end in a dystopian dictatorship beyond any of the worst nightmares of people like Orwell, or else in a Frank Hebert's "Butlerian Jihad" is something that I am at a loss of imagination to see.
I've never heard anybody bemoaning this development - agricultural field jobs are back-breaking and dull - but it did create a situation of greater societal dependence on a much smaller number of farmers. Ensuring that everyone gets enough food then becomes something of a socioeconomic problem rather than a technical problem.
Now, if LLM-related technology starts making major inroads into the well-paid upper-tier employment sector, this could get a bit more interesting. My favorite suggestion is to use LLMs to replace shareholder-appointed corporate board members - and hey, why not the shareholders themselves? LLMs might be much better at generating capital from investments, and as they're not wasting the money on giant yachts and private islands, they might be much more efficient at plowing capital gains back into research and development and so improving manufacturing.
This sort of highlights the fact that a society in which all the wealth is in the hands of a few aristocrats is a society doomed to eventual extinction, particularly in an era of rapid technological innovation.
I think it's probably because we're constantly bombarded with mainstream narratives and people in our peer group are shown the same news articles and content at the same time. People who actually believe the articles/mainstream narratives tend to come across (to the rest of us) as though they've been programmed because they're all saying and thinking about the same thing at the same time in perfect unison. There's something eery about that.
Associating your human worth with your economic productivity and then limiting exponential economic productivity to a few already powerful people is a nightmare.
Economies of scale can be rephrased as winner take all.
Never thought I'd say this, but I am glad post modernism exists. A philosophical movement that rejects grounded representations of art or philosophy is somehow the perfect bulwark against AI.
I imagine a future with few jobs, where everyone is largely self sufficient. Would that really be so bad?
People who own nothing that can produce continuous value will not have anything to trade with in the long term and will effectively be excluded from the economy.
People who don't want to depend on the government could always spin up their own AGI and become self sufficient. Some solar panels, a few robots and some farmable land are probably all you need when it comes down to it.
People who own the means of production, the factories, the mines, the farms, the power stations, … etc. will control who they sell their goods to and for how much.
Having to rely on the government is scary. If you got a bad government you are so screwed. You have nigh zero leverage as you have nothing to offer anyone.
news.ycombinator.com##td.title:nth-of-type(3):has-text("AI")