This post makes me feel a bit nervous about what kind of behavior might stem from holding these mindsets around abundance and scarcity - especially if the gist of this is misguided (in practice). I'm interested to read the next post.
I think some of the conclusions around energy are missing some super important context that may lead to opposite conclusions. I think Nate Hagens is someone who describes this perspective quite well. A couple points that feel relevant: Much of the trend of energy getting cheaper is that we were historically getting better at extracting fossil fuels, especially when we use fossil fuels to get more fossil fuels. We're seeing diminishing returns, and the 'abundantly cheap' energy we see now is an anomaly as we draw down resources which took millennia of sunlight to accrue. A second point is around renewable energy - the challenge comes with storing energy efficiently. With current battery technology, it doesn't seem like there are enough rare earth metals to make this a viable solution for large scale implementation with our current energy usage habits (Peter Zaihan makes this case well).
In my part of the network, we come from an understanding that the physical world contains natural scarcity, and that the abundance comes in our subjective experience - things like love, beauty, joy, and meaning can become decoupled from material consumption (to a certain extent), and that this is a safer path for humanity to take, rather than to treat energy and materials as abundant and aligned-AGI as the cure-all.
That said, I'm looking forward to the next piece and where you take this train of thought - diversity of perspectives is the best way to understand a complex topic like this one! And I'm also happy to schedule a chat if you'd like to compare notes :)
Batteries don't need rare earth metals. You might think of cobalt, but that is not part of the "rare earth" metals, and many battery chemistries don't even need that any more.
What currently does need them are the permanent magnets in electric motors, but if you have more abundant energy they can be replaced with electromagnets.
There's lots of options for energy storage. The main problem is one of pricing. Add a carbon tax to the mix until it's cheaper to build a wind farm *and* a battery farm instead of a gas boiler feeding a generator, and that'll change.
I will have a deep dive on the infinite energy topic later this year—hopefully.
The gist of the idea, from what I've seen, is that what Matthias is describing is always what ends up happening. We think there's a physical limit until it turns out there isn't, we just didn't know.
The magnets are a good example.
Another one is chemical energy storage. You can craft natural gas out of CO2 and electricity, and use the current NatGas infra at no additional cost. The only thing needed here is a drop in cost per kwh of solar elecricity, which is happening. You wouldn't need batteries if you can store your energy as NatGas you burn and recycle from the air.
Naryan, you might not agree with my next article, but it sure will be provocative!
Well, if we really want to use the current NatGas infrastructure, we also need to get the CO2 to build that gas out of the air. Less than half a percent isn't much, so you need much more new hardware than converting as many sectors to electricity and/or closed NatGas+O2 systems as you possibly can.
Yes an utopian free energy and free intelligence world is something we should strive to achieve IMHO, but the problem is getting there.
I don’t think scarcity is the main cause of wars. There aren’t many wars that I can think of where scarcity is the main cause—although there are a lot of new arguments that various eras of climate change drove some—like for example the 17th C. But, even then, the stated causus belli was almost always ideological. Once upon a time, most wars were about what we would call glory and plunder. Then many were about religion or ideas of human liberty and dignity. Wars continue now in times of relative abundance—Ukraine is not a war about scarcity, nor is the Syrian conflict. The Libyan Civil War is not about scarcity, though competition over abundant resources are central to it, but about who will rule. AI might make ideological wars more vicious. Anyways, my small quibble—interesting article, thanks.
I think the question is: What % of violence disappears if you eliminate material scarcity.
There are 2 types of violence, internal and international.
On internal violence, I think we agree that a huge amount of violence is due to scarcity (linked to theft, or control of resources).
Internationally, you have things like lebensraum (WWII, even if there was an ideology behind), the oil wars in the Middle East, all colonial and post-colonial wars, most wars in modern and pre-modern wars were indeed about plunder and expanding the lands controlled (even if glory was a component, it was linked to how many resources were controlled)...
So you're right that this won't kill all wars, but it should reduce them proportionally to the amount that is caused by scarcity, which I would subjectively estimate at 70% of casualties. If there's an analysis on this, I'd be curious!
To be nitpicky: I think your definition of scarcity is awfully broad. If we are talking about material scarcity with respect to the cause of wars as opposed to all violence per se, most wars are not because of want, but because of ambition, which may be about trying to control more resources, but not because they lack them outside of the fact that they don't control all of them. Japan's imperial wars might fall under the category of material scarcity, but Pearl Harbor is of a different kind of calculation. Lebensraum may have been an official goal of NAZI policy, but Germany faced no real material scarcity: people were not starving, water was plentiful, factories had enough inputs for all manner of goods, electricity was general across Bavaria, there were no serious shortages in 1939. Napoleon didn't set out to conquer Europe because people needed to be fed back in France, but rather in spite of it.
Alexander the Great wasn't particularly worried about material scarcity in Greece that needed to be rectified, he wanted to be king of the world. No material scarcities back in Macedonia were satisfied by the conquest of Afghanistan. There are kinds of scarcity which affect the calculations of people who start wars, but material scarcity is rarely one of them, unless you mean to say that for some people nothing is enough, so for them there is always scarcity.
Pearl Harbor was a pre-emptive strike to the US when the US threatened to block Japanese access to oil, which would have created the worst type of scarcity to a growing empire.
Germany didn't have access to oil either, and based on the farming techniques at the time, land productivity was getting close to maximums. There's a reason why the Haber-Bosch process for fertilizer was invented by Germans.
Napoleon set to conquer Europe because the rest of Europe was a monarchy, and they couldn't accept a Republic to succeed, for fear of losing the grip on their own power. So all monarchies threatened France, and Napoleon was just more successful.
All of this to say that it's not either or. Both components are more or less important depending on the case, and I believe (without substantiation, would love to see data) that scarcity causes more than half of violence.
If we looked back at the last 100 years, could we discover how much of each type of energy was used in creating the population each year? Clearly solar and nuclear energy would be insignificant in 1945, and the vast majority of humans would be the result of burning coal and oil.
The point is that every time there is a technological revolution (such as agriculture / coal / oil etc) there is a leap in the size of the human population. It seems likely that the next leap will happen with very low cost fusion energy.
I find it amusing to suggest that we have simply converted oil into humans.
Amazing article as usual. Can u spend some time on the AGI alignment problem. Lex freedman had an episode with Eliezer Yudkowsky recently and it’s down to a very simple premise: creating a system that is more intelligent than humans is pretty bad idea with catastrophic outcomes: demise of our species. In the ensuing discussions people attack the person but not his logic and argument. It’s the usual sad state of Twitter discussions. But your at length expositions are able to dig deep and reveal subtle nuances and shed useful light on complex topics. In this article you are talking about the positives of super intelligence. Worth exploring the potential risks ?
I must confess that I was tempted to “cheat” and go straight onto reading the next article in the series, but I seem to have managed to resist temptation. I’m sticking with the idea that it is more useful to try to clarify my thoughts before reading yours, particularly as we are now getting to the heart of the matter.
Why does food cost money? Because it is OWNED by someone else.
Why are there people who can’t afford it? Because they don’t OWN enough wealth to buy it
Why is food scarce? Overall it isn’t these days. The distribution of the world’s wealth and power prevents it from getting to those who don’t have enough.
Land useful for humans is somewhat scarce relative to the world’s population which is why it has been such a valuable asset and such a good store of wealth if you own it.
Energy is not at all scarce, but certainly is in its more useful, concentrated forms. Owning the rights to extract oil and coal have been extremely good ways of accumulating wealth.
Yes, scarcity is mutable. So if you produce an abundance of something it becomes less financially valuable and instead of helping someone acquire wealth, it can make them lose it. Business people know this: they carefully control production to avoid excessive supply and consequent price reduction. But of course you have to also reduce competition or someone else will undercut you. So the barriers to reducing scarcity often come from those who are going to lose out.
Knowledge used to be a precious commodity. But with the advent of the internet, people often now give it away free of charge!! It still has immense value, but financially it has been devalued. The supply of computational power has likewise increased dramatically and its cost plummeted. Intelligence has also increased. Why? More people with more access to information and to each other, plus a healthy dose of competition.
Has wisdom increased? I would vote no, in fact I think overall it has decreased. This means that the intelligence to wisdom ratio has become very unbalanced, a potentially dangerous state of affairs when powerful destructive weapons are available. Why do we even need wisdom? Because people all have different needs and wants which sometimes come into conflict. They are programmed for conflict and competition to varying degrees, so ideally we have to work out safe ways of resolving these conflicts. I agree that reducing material scarcity will help greatly, but there are human concepts which cannot be created because their very nature is relative (e.g. wealth, power, social status, mojo*). If they were not scarce, they would lose their meaning. Competition for these limited resources can lead to conflict. Is it possible to create Artificial Wisdom? Maybe. It might be more acceptable to some people coming from a computer than from humans who have distilled it from thousands of years of human experience.
*I still haven’t come up with a clear definition of what I mean by mojo but will keep trying
This was great as always. Could you do a post about your writing, research, and reading process? Some of this is obviously due to great skill in storytelling but I think there’s still a lot that others could learn from how you do things. Or in Tyler Cowen’s words: What is the Tomas Pueyo production function?
Thank you for helping expand my understanding of the economy and envission future possibilities. At the same time I find it hard to share your optmism, since I see that each decade more wealth is concentrating in fewer hands. The rich are getting richer, the poor are getting poorer and the environment is deteriorating really fast. I don't see that, under capitalism, technological developments really brings solutions to the main problems of society
As you can imagine given my previous articles, I agree with the fear.
But this is a different problem. This is a problem *before we have infinite intelligence*. Once we do, there isn't scarcity anymore, so inequality doesn't exist either.
Thank you. It's difficult for me to foresee what would all the implications of having infinite intelligence be. For now what I can see is that the richest people in the world have more money then they would ever be able to spend, and a small fraction of they''re income would be enough to pay for everybody in the planet to have basic food, water and shelter
Actually, that's not true. We're not at that stage yet. The richest 1% of Americans gained $6.5T in wealth last year.
That's a big number!
But even only in the US (so not the rest of the world), that's a little less than $20K/American. Which is hard to survive on in the US. And that's if you confiscate _all_ the created wealth of the top 1%, which obviously would have a negative impact on future wealth creation. And 2021 was a good year. Some years, they lose trillions, in which case, how do you feed/clothe/house everyone?
People have to realize that we are still definitely not in a post-scarcity world.
Solves scarcity, but, to the point of your answer to the previous comments, there is a big, frightening, empty space...
Solving scarcity in land, materials and energy would generate oversupply of human capacity to work, or otherwise stated, of humans.
Can illimited intelligence integrate the need to solve wealth (if still a useful unit of measure) redistribution amongst "useless" humans? And can it do it without violence, forced elimination, and in a time framework corresponding to its advances in the other fields?
The three laws of robotics come to mind, and I am really interested in what comes out of the 6 months moratorium proposed by Musk & others.
Your post of today brings me back to one of your first non-covid publications. My fears were the same. I keep on reading all your publications with the same interest. Congratulations!
Yes, inequality is a very troubling problem today, in the Pre-Infinite Age. Well, let's be more precise: it's troubling to *some people.* It's certainly not something that the winners of capitalism are especially troubled by, beyond the crocodile tears from the Davos set. It's certainly not a dynamic that the people benefiting from are willing to change, from the wealthiest 1% on down to the merely comfortable "Dream Hoarders" in the upper-middle-class 20%. We could today feed the world with calories to spare right now, but we don't. We could redistribute wealth and income to satisfy at least a $10,0000 middle-income lifestyle for every one of the Earth's 8 billion people, but we don't. Seems an intolerable downgrade to those of us with more, even though it would fund a bounty that would make aristocrats of yore blush. There's really no upward limit on our "needs," when an American making six-figures dreams of becoming a millionaire and a millionaire dreams of billions. Being wealthy and high-status is more satisfying than being good and sharing, clearly. Don't you agree? Yes, YOU. Because you are certainly part of that group of relatively wealthy Dream Hoarders, aren't you? I am. And my progressive ethical or political beliefs, notwithstanding, when it comes down to it, I'm complacent to live in an unequal society. I'm happy enough to pay a hefty share of taxes on my income to fund transfers to others and to fund universal public services, sure ...but I'm certainly not giving everything I don't strictly need away to people who need it more. So, I'm part of the problem. We can justify it all day long, but we're just being unethically selfish by not sharing more in a world full of want. Not that you can really blame us for just being selfish, like most everyone else. David Graeber and David Wengrow's "The Dawn of Everything" thesis notwithstanding, it seems that inequality and elite efforts to maintain it are about the most consistent dynamic in the history of human civilization. Plus, exposure to wealth and high-status is proven to actually reduce our capacity for empathy, so, if "sharing is caring," then we, the relatively wealthy ones, have become uncaring in our comfort.
So, if you haven't gotten any satisfying answers to inequality, the (unsatisfying) answer is that wealthy and powerful people enjoy having more and that any solutions to redistribution feel like losing when you're on top. "Yes, but that's a malady of the Old World of scarcity," you're saying, "There cannot be inequality or want in the Infinite Age following the Singularity!" Infinity will render us ethical...
But WHY? What's the mechanism for change? Infinite intelligence can be a paradigm shift, but it is also force-multiplier or accelerant for existing conditions. Previous technological revolutions have tended to merely crystalize historical contingencies. It's a pretty arbitrary historical condition that the primary agents and beneficiaries of the computing age were white American men who had the means to study at places like Stanford (set up by a rather nutty wealthy heiress with railroad money... showing how the gains of previous technological bonanzas are laundered into the present). Were these West Coast dudes the smartest minds on the planet from the 1970s? Or even the best and the brightest from within the entire United States. They probably think so, and a lot of their fans agree, but the premise is absurd. They were certainly very clever young men, all... but standing on the shoulders of the existing power structure. And the other clever young men and women everywhere else weren't.
Conversely, historical dynamics don't let go of the people with less wealth and power, even after we have cool iPhones and ChatGPT. Americans argue to death why their Black citizens tend to have such stubbornly inferior lifestyle outcomes compared to white and Asian Americans. But isn't the answer as simple as: The former group endured 400 years of slavery, followed by another century-plus of explicit, legal, violent discrimination, and since then a few generations of more implicit structural racism. If that's not the reason, then it's a startling coincidence. This "natural experiment" in wildly divergent outcomes within the same society becomes even more clear when you look to the less remarked upon but even more terrible life outcomes of Native Americans, who never recovered from the historical sin of colonialism and genocide. No innovation dreamed up by Americans has changed this initial condition. Maybe because we can't. Maybe because we just don't care enough to try. And, though intentional advances against other aspects of inequality have been more successful, you still see the unmistakable mark of social conditions that existed four centuries ago in our country, despite all the creative destruction since. That's how path dependent inequality is. And it's not just the case in the US. In Italy, you can trace the wealth of the same wealthiest Florentine families all the way back to the Medici Era. In Sweden, that paragon of equality, the same eight families have dominated commerce and politics for centuries, in most cases back to Sweden's 16th Century independence and earlier!
So, why should an era affected by AI be any different? Our experience with AI certainly hasn't been encouraging so far. Building these models is the same group of American men (American IT workers are actually even more male than they were in the 1980s!), now joined by the imported elites of India and China. We've already seen that their models reproduce the attitudes and biases of the current power structure. AI still proves itself to be embarrassingly racist and sexist, which isn't at all surprising considering that it learns from existing human patterns. So, what mechanism do you imagine would make a more developed and capable AI something that would serve the interests of the many, rather than the few? After all, follow the money. Who owns AI today? What is it for? *Whom is it for?* Even the deceptively named "OpenAI" is owned by... Microsoft. It was a startup. But not a scrappy one. It was created by the same "PayPal Mafia" gang of Elon Musk, Peter Theil, et al. The ghosts of Internet 1.0 emerging again. And productized so far to drive Microsoft's legacy ad-driven search and monopolistic productivity suite businesses (and undoubtedly some national defense applications). The only other viable AI competitors today are from the same FAANGs who have gathered up all the treasures of the Internet 2.0 Era. As they have for two decades, these few companies deepen their "moats" by buying up any upstart rivals. Not that there's much danger of that. It's near-impossible to create an AI in your garage, anyway: AI requires tremendous resources--in data, computing, and training--to be viable. Network effects reign. Regular people will never be able to compete in this arms race. And, so far, no (democratically-controlled) state is stepping in to decentralize control, either. Even in "Communist" China, AI is conceived as a tool for building the national wealth and power. Not for making individual Chinese peoples' lives better. The Means of (Infinite) Production will everywhere be even more firmly in elite hands.
Just as with AI's precursors and antecedents--telecommunications and computing--AI will inevitably centralize wealth and power. In the same way as the original Gilded Age was made possible by how the Second Industrial Revolution undermined skilled labor interests, computing technologies and automation have made inequality *more* viable in this, our New Gilded Age. You just don't need as much labor, skilled or otherwise, to make a marginal unit of GDP today. So, you now get a society split into the 20% and the 80%. And that will seem tame, equitable, and "inefficient" compared to the Third Gilded Age of the Mid 21st Century. When the AI Wizards are conducting transmutation alchemy with limitless fusion energy, who needs labor or input from the poors at all! Or even the (formerly) normie middle classes, tapping away in their "bullshit jobs?" Just because we are freed from resource scarcity doesn't mean that there's any need to actually share those infinite resources. It will be the gods quibbling on Mount Olympus and the rest of us underfoot.
Would be great if we actually see quantification of that claim / slogan about rich getting richer but mostly about the poor getting poorer. Last I checked China lifted 800m out of poverty in a single generation.
Very interesting essay. But one quibble which is actually pretty major for undermining (pun!) your thesis: raw materials are (and will be) still scarce, even with the input of all the intelligence imaginable.
You said, "This is why peak oil has been declared for decades, but in fact we might well find peak oil not because we can’t find it anymore, but because we don’t need it anymore." Firstly, we have arguably already hit Peak Oil, in terms of peaking oil production. The IEA called for this year already. And, yes, that prognosis has been a slippery one, but we have some accumulating facts that point in the same direction: Conventional oil *discoveries* peaked in the 1960s. Conventional oil *production* peaked more recently, in the 2010s. Per capita, production's been going down since 1979, as population growth outstripped production increases. As for "proven" oil reserves... well, that's a very contested matter involving a lot of corporate and national secrecy and perverse incentives to overstate things. But, in any case, we're not actually pulling more oil out of those proven reserves, however large or small they may be.
Does that mean the oil ran out? No. There's still oil being produced, obviously. And there will be oil produced for some decades more, presumably. But the limiting factor has always been: Can a marginal unit of oil be can be extracted *economically* at a given price? If oil's $60/barrel, that reduces significantly the amount of proven reserves that would actually be worth tapping. And, at a certain point, unconventional sources are so energy-intensive, complex, and expensive to tap, they aren't reserves at all, anymore. Even at $120 or $200/barrel. Long after we stop producing oil, there will still be oil under the deepest parts of the ocean floor or aggregated with a much of non-useful gunk under the ground somewhere in Canada. But there it will remain, forever be too much trouble to extract, just like mining on an asteroid is likely to be.
And even if you take finances out of it, an even more limiting factor on whether an oil barrel goes from notional to actual is the Energy Return on Investment (EROI), or the energy input required to extract a marginal unit of energy output. And that's where innovation hasn't fixed our issue. Shale oil production, a significant innovation drawing from impressive amounts of human intelligence and skill, resulted in more oil production (especially in the United States), but the EROI decreased near to the point where it is costing nearly as much energy to get the stuff as it produces. After the initial rush of optimism and triumphalism, many of those unconventional producers quickly found that shale oil was never actually profitable, and was only feasible with a lot of easy financing. But it's not just shale oil running into this limit. Fracking shook loose a lot more methane, but at a cost. Getting the natural gas out of the ground and transporting it across oceans as LNG the way we're doing much more now since Russia invaded Ukraine makes that methane far more expensive. And, by extension, it makes energy too expensive to use, beyond a certain point. You see this with much of the German industrial base just closing up shop temporarily or permanently, given the expense of industrial heat from burning methane that now must be sourced as LNG, instead of piped from Russia. And, with oil, as we've already tapped 2/3 of discovered fields, see productivity of our most prolific fields trending in the wrong direction, and scrounge about for ever more marginal and harder-to-access discoveries, we'll be reaching the point where the EROI of fossil fuel energy production, overall, approaches 1. And that's very bad. One unit of energy to get one unit of energy is very different from the exponential effect that fossil fuels had for human civilization in the 19th and 20th Centuries, back when the ratio was closer to 100:1. A hundred years ago, you could be as inefficient with the stuff as you like because it was so cheap! Today, even with very mature fossil fuel technologies that approach the limits of efficiency, we can't get much more useful energy out of a barrel of oil that keeps getting scarcer.
But is this just because we're past the age of oil, and we "don't need it" anymore? Hardly! Oil demand fluctuates and there have been impressive gains made in replacing oil as an input for electricity production or passenger transport, but we're still very much in the oil age. Oil is still the only primary energy medium which is so energy-dense, stable, transportable, and versatile. We use fossil fuels for EVERYTHING. Including the manufacturing of the solar panels on my roof and the lithium-ion battery that helps my hybrid vehicle to go. We even basically eat the stuff, as you alluded to in your aside about fertilizers.
Oh, but we'll have a solution for that, too, you say. Though you didn't mention it under the Raw Materials section, perhaps nuclear fusion's limitless energy could allow for the transmutation of other feedstock into oil or other synthetic fuels? Perhaps we could create endless hydrogen? Perhaps, but this is relying on a kind of deus ex machina, isn't it? Viable fission doesn't even exist in prototype form yet, much less scaled out commercially over the whole face of the earth. And not for a lack of a half-century of trying. That transition, if it happens, won't be something for the first half of this century, if we see it this century at all. And aren't even the unimaginably complex machines that we hope will create fission manufactured with those same scarce inputs? You don't get a spherical tokamak fushion reactor without a hell of a lot of high-spec metals, specialized magnets, and high-voltage electricity. And you don't get those without the fossil fuel-driven processes that allow us to mine things, apply industrial heat to them, and manufacture and assemble complex machines to unbelievable exactitude.
So, I don't know quite techno-optimists think that nuclear fusion will be so easy and cost-free in the way that no other technology before it has been (remember that nuclear fission was also expected to be "too cheap to meter"). The costs in the inputs of financing and scarce resources will be largely front-ended, just like with nuclear fission, but they will be significant. And given the progress against this technology so far, we will be trying to build hundreds or thousands of fusion plants after 2050 just to keep up with our conventional energy needs, the time by which we're likely in the thick of a "polycrisis" of dwindling fossil fuels , essential resources (like phosphorus) that are also exhausted, energy that is expensive and unreliable, food production that is insufficient, all the various other knock-on effects of Climate Change, and all the political and geopolitical effects thereof.
The entire thesis of today's article is very much that an infinitely intelligent AI is a Deus Ex Machina. It can solve problems in ways we can't even foresee.
That said, I'll give you a path *we can already foresee* for energy.
Drop photovoltaic costs enough (as it's clear it's happening) and energy soon becomes too cheap to meter. With that energy, you can generate all the fossil fuels you want just by extracting CO2 from the atmosphere, in a cheaper way than pumping them out of the Earth. Another one you can easily do is drop nuclear fission energy costs tremendously, because it can simplify the designs while improving them, in a way that they're much safer than any other energy source (already true today) and also cheaper (not true, but because of a limit in intelligence).
So infinite intelligence does solve the problem of not enough energy or fossil fuels.
The transmutation argument is different. It's not for the scarcity of molecules (simple to solve with infinite energy), but for the scarcity of elements (like Bismuth or something).
I don't think it's so hard to imagine infinite intelligence solving the issue that there are trillions of Suns already, but we just don't know how to make them and use them. But if you can help me understand what's the actual physical barrier to that, I'd love to hear!
It's a common working assumption that some form of AI now in development, including ML, LLMs, and other forms of artificial intelligences not yet conceived of--will at some point hit that "Singularity" and then beyond to an exponential age of intelligence that makes Moore's Law look trite.
Maybe.
But, again, AI has been in development about as long and gone through even more hype-cycles than nuclear fusion. Skepticism for both is warranted. Exponentiality tends to run out of steam. Especially when the initial conditions for progress don't continue indefinitely (more on that later).
But let's assume that we can and do supplant the Intelligent Design God of the 18th Century imaginings, rearranging the very molecules of the Universe in ways that are most useful to us. Still, even there, gods and other magnificent beings run into limits. The God Yahweh had limits. After six days of Creation, He had to rest, according to Genesis. And that was just the beginning of all the frustrations with His Creation! There's book(s) about it. Very popular reads. But, look, too, at the more empirical facts of our existence: the Sun itself has limits! It creates energy "for free" insofar as it self-immolates and will expire in 5 billion years or so (which isn't as relatively long-seeming in Sun terms). It will start to expand outward as it ages, like so many of us, losing its youthful heat and vigor in the process as it burns through initial endowments. Even the wider Universe has limits! As it ages, it also expands and cools (we think), which leads inexorably toward the Heat Death of the Universe (presumably). That's a bummer for any Demiurge behind it all because even information itself may not be able survive that (we think). The final cost of the Big Bang? All that doing smart stuff to just be forgotten!
I raise these extremities to make a point, but pull the bounds closer to our scale and you see that there are limits here, too. Everything we use now was created via incredibly powerful celestial or geological forces beyond both our agency and ken, burning through unfathomable amounts of the ultimate scarce and non-renewable resource: TIME! Given the evolutionary nature of the Universe, which has operated over a 13-14 billion years of that ultimate scarcity creating practically infinite variations of matter and activity, isn't it a bit hubristic to think that we can recreate the magnitude of that power without even as much time involved?
Could we use super-intelligence to turn a mountain into a freshwater lake? Well, the Earth did it all the time, but it took a few million years. Could we turn all the Crocs that more primitive humans created in the 2010s into something non-renewable like helium so that we can have funny voices and asthma sprays forever? Well, the Universe kinda basically did that, but it did take a few billions years. So could we accomplish the same in a few hours? With enough energy; from enough intelligence? Maybe. Energy is all around us: There are an estimated 47 TW of heat energy flowing from the Earth's core. Even more energy, 173,000 TW of solar energy hits the Earth from the Sun. The rest energy embedded in all the matter on Earth is incalculable. This is A LOT of juice! But, remember, that energy does something already. A lot of it is not available for human consumption. Or at least not without displacing all the other "ecosystem services" that exist outside of our direct interest, which we have done in so many other aspects of life (e.g. displacing almost all large terrestrial mammals, to make room for our meat or blowing through nearly all the arable land on the planet for our cereals).
So, even if it's possible to capture and apply all this energy, what's the equal-and-opposite reaction? What's the unforeseen externality? What's the cost? If we haven't been intelligent enough to look out in front of our own cleverness to foresee the consequences of our own cleverness, will an AI? Why do we assume that an AI would be "wiser" and more long-termist than us, rather than just as monomaniacal and instrumental in thinking as we have been? Carpet the entire planet with super-efficient solar panels and drill boreholes to suck up the last residual heat we can get from the Earth, in order to put on the real show of rearranging the entire Solar System for post-scarcity profit! Because there is never "enough." If there were, why would we have billionaires today?
This post-scarcity bonanza could also inevitably result in wars on an unimaginable scale. Because your summation for the reasons for war (and their obsolescence) could be naïve. Any good Space Opera will remind you that "scarcity" is always relative and more infinity is always better than less. Even the utopian society of Star Trek has fighting. So maybe we'd (or rather the AI versions of us) would just be wringing the last available units of energy from our Earth and its Sun to power weapons. Weapons that would become their own justification as adversary intelligences also matched them with weapons of equal and greater potency. A Universe-wide arms race hoovering up all the matter and energy until... well, something terrible that fast-forwards the timeline to the Heat Death of the Universe for the expiration of the intelligences that the Universe sprang forward to know itself.
But, again, that's *if* we (or the aliens) could even pull any of this off. It's a good bet that we also just will never be able to. Or that an intelligent species like ours could, in theory, but always ends up destroying themselves with other means long before (Fermi Paradox). Circling back to my point about the timeline of fusion: What if we *could* make it happen, but won't ever quite get there because our other scarcities catch up with us first? We hit a wall from 2030-2050, where there's too much instability and increasing scarcity in the necessary resources to finish the work of commercializing and scaling out fusion, AI, or other exponential technologies? Could we find ourselves like a race of Nikola Teslas, full of genius ideas never quite fully realized in a too-short life ending in frustrations and penury?
If we sum up what you say, there are two limits to scarcity:
1. Physics
2. Non-physical stuff, like for example things tied to humans (love, sex with a human, social hierarchy)
We agree on that.
But I think physical limits are extremely far from anything imaginable today, so compared to the world today, we will reach many physical infinities. The same way as today heat, electricity, water, and computing power are near infinite for you, so can most other physical limits be after intelligence becomes infinite.
Then there's point number 2, which I explore in the article this Thursday
I started a company last year that uses data analytics (our AI still being built) to make land use and human work more efficient for the deployment of renewables (won’t name it as I am anon here). Your outline is a more all encompassing description of what we talk about in some of our longer product meetings, cool article!
Re : scarcity of ressources. Don’t forget asteroid mining. With fusion propulsion it shouldn’t be a problem. Will bring a *massive* amount of raw material
This post makes me feel a bit nervous about what kind of behavior might stem from holding these mindsets around abundance and scarcity - especially if the gist of this is misguided (in practice). I'm interested to read the next post.
I think some of the conclusions around energy are missing some super important context that may lead to opposite conclusions. I think Nate Hagens is someone who describes this perspective quite well. A couple points that feel relevant: Much of the trend of energy getting cheaper is that we were historically getting better at extracting fossil fuels, especially when we use fossil fuels to get more fossil fuels. We're seeing diminishing returns, and the 'abundantly cheap' energy we see now is an anomaly as we draw down resources which took millennia of sunlight to accrue. A second point is around renewable energy - the challenge comes with storing energy efficiently. With current battery technology, it doesn't seem like there are enough rare earth metals to make this a viable solution for large scale implementation with our current energy usage habits (Peter Zaihan makes this case well).
In my part of the network, we come from an understanding that the physical world contains natural scarcity, and that the abundance comes in our subjective experience - things like love, beauty, joy, and meaning can become decoupled from material consumption (to a certain extent), and that this is a safer path for humanity to take, rather than to treat energy and materials as abundant and aligned-AGI as the cure-all.
That said, I'm looking forward to the next piece and where you take this train of thought - diversity of perspectives is the best way to understand a complex topic like this one! And I'm also happy to schedule a chat if you'd like to compare notes :)
Batteries don't need rare earth metals. You might think of cobalt, but that is not part of the "rare earth" metals, and many battery chemistries don't even need that any more.
What currently does need them are the permanent magnets in electric motors, but if you have more abundant energy they can be replaced with electromagnets.
There's lots of options for energy storage. The main problem is one of pricing. Add a carbon tax to the mix until it's cheaper to build a wind farm *and* a battery farm instead of a gas boiler feeding a generator, and that'll change.
I will have a deep dive on the infinite energy topic later this year—hopefully.
The gist of the idea, from what I've seen, is that what Matthias is describing is always what ends up happening. We think there's a physical limit until it turns out there isn't, we just didn't know.
The magnets are a good example.
Another one is chemical energy storage. You can craft natural gas out of CO2 and electricity, and use the current NatGas infra at no additional cost. The only thing needed here is a drop in cost per kwh of solar elecricity, which is happening. You wouldn't need batteries if you can store your energy as NatGas you burn and recycle from the air.
Naryan, you might not agree with my next article, but it sure will be provocative!
Energy of course is not scarce at all. It just doesn't exist in the form or concentration that is useful to us
Well, if we really want to use the current NatGas infrastructure, we also need to get the CO2 to build that gas out of the air. Less than half a percent isn't much, so you need much more new hardware than converting as many sectors to electricity and/or closed NatGas+O2 systems as you possibly can.
Yes an utopian free energy and free intelligence world is something we should strive to achieve IMHO, but the problem is getting there.
Stay tuned!
I didn't understand this at all. So I fed it into Bard and asked for a summary. Bard was equally confused.
I mean, sure, in theory, the solar system (and universe) is finite. In practice, we're going to be long dead before we run out of energy and matter.
We'll probably see asteroid mining within my lifetime.
I don't think we will! But you'll have to wait to the next space article for that...
I don’t think scarcity is the main cause of wars. There aren’t many wars that I can think of where scarcity is the main cause—although there are a lot of new arguments that various eras of climate change drove some—like for example the 17th C. But, even then, the stated causus belli was almost always ideological. Once upon a time, most wars were about what we would call glory and plunder. Then many were about religion or ideas of human liberty and dignity. Wars continue now in times of relative abundance—Ukraine is not a war about scarcity, nor is the Syrian conflict. The Libyan Civil War is not about scarcity, though competition over abundant resources are central to it, but about who will rule. AI might make ideological wars more vicious. Anyways, my small quibble—interesting article, thanks.
I think the question is: What % of violence disappears if you eliminate material scarcity.
There are 2 types of violence, internal and international.
On internal violence, I think we agree that a huge amount of violence is due to scarcity (linked to theft, or control of resources).
Internationally, you have things like lebensraum (WWII, even if there was an ideology behind), the oil wars in the Middle East, all colonial and post-colonial wars, most wars in modern and pre-modern wars were indeed about plunder and expanding the lands controlled (even if glory was a component, it was linked to how many resources were controlled)...
So you're right that this won't kill all wars, but it should reduce them proportionally to the amount that is caused by scarcity, which I would subjectively estimate at 70% of casualties. If there's an analysis on this, I'd be curious!
To be nitpicky: I think your definition of scarcity is awfully broad. If we are talking about material scarcity with respect to the cause of wars as opposed to all violence per se, most wars are not because of want, but because of ambition, which may be about trying to control more resources, but not because they lack them outside of the fact that they don't control all of them. Japan's imperial wars might fall under the category of material scarcity, but Pearl Harbor is of a different kind of calculation. Lebensraum may have been an official goal of NAZI policy, but Germany faced no real material scarcity: people were not starving, water was plentiful, factories had enough inputs for all manner of goods, electricity was general across Bavaria, there were no serious shortages in 1939. Napoleon didn't set out to conquer Europe because people needed to be fed back in France, but rather in spite of it.
Alexander the Great wasn't particularly worried about material scarcity in Greece that needed to be rectified, he wanted to be king of the world. No material scarcities back in Macedonia were satisfied by the conquest of Afghanistan. There are kinds of scarcity which affect the calculations of people who start wars, but material scarcity is rarely one of them, unless you mean to say that for some people nothing is enough, so for them there is always scarcity.
Pearl Harbor was a pre-emptive strike to the US when the US threatened to block Japanese access to oil, which would have created the worst type of scarcity to a growing empire.
Germany didn't have access to oil either, and based on the farming techniques at the time, land productivity was getting close to maximums. There's a reason why the Haber-Bosch process for fertilizer was invented by Germans.
Napoleon set to conquer Europe because the rest of Europe was a monarchy, and they couldn't accept a Republic to succeed, for fear of losing the grip on their own power. So all monarchies threatened France, and Napoleon was just more successful.
All of this to say that it's not either or. Both components are more or less important depending on the case, and I believe (without substantiation, would love to see data) that scarcity causes more than half of violence.
You might be interested in my new post
I agree. AI will alleviate material scarcity, but may exacerbate other issues (humans with superintelligent AI fighting over status, for instance).
It could be argued that the population of the planet today is largely the result of oil. That is to say; we convert oil into people.
Energy!
Hunting/gathering > Agriculture > Coal > Oil > Solar/Nuclear
If we looked back at the last 100 years, could we discover how much of each type of energy was used in creating the population each year? Clearly solar and nuclear energy would be insignificant in 1945, and the vast majority of humans would be the result of burning coal and oil.
The planet of today won't be the planet of the future, just as the planet of 1945 wasn't the planet of 1845.
The point is that every time there is a technological revolution (such as agriculture / coal / oil etc) there is a leap in the size of the human population. It seems likely that the next leap will happen with very low cost fusion energy.
I find it amusing to suggest that we have simply converted oil into humans.
That was all before cheap contraception, though. So I don't expect a big leap in human population with another energy revolution.
Infinite intelligence solves this!
Amazing article as usual. Can u spend some time on the AGI alignment problem. Lex freedman had an episode with Eliezer Yudkowsky recently and it’s down to a very simple premise: creating a system that is more intelligent than humans is pretty bad idea with catastrophic outcomes: demise of our species. In the ensuing discussions people attack the person but not his logic and argument. It’s the usual sad state of Twitter discussions. But your at length expositions are able to dig deep and reveal subtle nuances and shed useful light on complex topics. In this article you are talking about the positives of super intelligence. Worth exploring the potential risks ?
I thought everything had been said about alignment, but apparently there is room for more. Thanks for this.
I must confess that I was tempted to “cheat” and go straight onto reading the next article in the series, but I seem to have managed to resist temptation. I’m sticking with the idea that it is more useful to try to clarify my thoughts before reading yours, particularly as we are now getting to the heart of the matter.
Why does food cost money? Because it is OWNED by someone else.
Why are there people who can’t afford it? Because they don’t OWN enough wealth to buy it
Why is food scarce? Overall it isn’t these days. The distribution of the world’s wealth and power prevents it from getting to those who don’t have enough.
Land useful for humans is somewhat scarce relative to the world’s population which is why it has been such a valuable asset and such a good store of wealth if you own it.
Energy is not at all scarce, but certainly is in its more useful, concentrated forms. Owning the rights to extract oil and coal have been extremely good ways of accumulating wealth.
Yes, scarcity is mutable. So if you produce an abundance of something it becomes less financially valuable and instead of helping someone acquire wealth, it can make them lose it. Business people know this: they carefully control production to avoid excessive supply and consequent price reduction. But of course you have to also reduce competition or someone else will undercut you. So the barriers to reducing scarcity often come from those who are going to lose out.
Knowledge used to be a precious commodity. But with the advent of the internet, people often now give it away free of charge!! It still has immense value, but financially it has been devalued. The supply of computational power has likewise increased dramatically and its cost plummeted. Intelligence has also increased. Why? More people with more access to information and to each other, plus a healthy dose of competition.
Has wisdom increased? I would vote no, in fact I think overall it has decreased. This means that the intelligence to wisdom ratio has become very unbalanced, a potentially dangerous state of affairs when powerful destructive weapons are available. Why do we even need wisdom? Because people all have different needs and wants which sometimes come into conflict. They are programmed for conflict and competition to varying degrees, so ideally we have to work out safe ways of resolving these conflicts. I agree that reducing material scarcity will help greatly, but there are human concepts which cannot be created because their very nature is relative (e.g. wealth, power, social status, mojo*). If they were not scarce, they would lose their meaning. Competition for these limited resources can lead to conflict. Is it possible to create Artificial Wisdom? Maybe. It might be more acceptable to some people coming from a computer than from humans who have distilled it from thousands of years of human experience.
*I still haven’t come up with a clear definition of what I mean by mojo but will keep trying
This was great as always. Could you do a post about your writing, research, and reading process? Some of this is obviously due to great skill in storytelling but I think there’s still a lot that others could learn from how you do things. Or in Tyler Cowen’s words: What is the Tomas Pueyo production function?
Thanks for sharing. Let me think about it
Thank you for helping expand my understanding of the economy and envission future possibilities. At the same time I find it hard to share your optmism, since I see that each decade more wealth is concentrating in fewer hands. The rich are getting richer, the poor are getting poorer and the environment is deteriorating really fast. I don't see that, under capitalism, technological developments really brings solutions to the main problems of society
As you can imagine given my previous articles, I agree with the fear.
But this is a different problem. This is a problem *before we have infinite intelligence*. Once we do, there isn't scarcity anymore, so inequality doesn't exist either.
In terms of material scarcity.
For non-material scarcity, see my next article
Thank you. It's difficult for me to foresee what would all the implications of having infinite intelligence be. For now what I can see is that the richest people in the world have more money then they would ever be able to spend, and a small fraction of they''re income would be enough to pay for everybody in the planet to have basic food, water and shelter
Actually, that's not true. We're not at that stage yet. The richest 1% of Americans gained $6.5T in wealth last year.
That's a big number!
But even only in the US (so not the rest of the world), that's a little less than $20K/American. Which is hard to survive on in the US. And that's if you confiscate _all_ the created wealth of the top 1%, which obviously would have a negative impact on future wealth creation. And 2021 was a good year. Some years, they lose trillions, in which case, how do you feed/clothe/house everyone?
People have to realize that we are still definitely not in a post-scarcity world.
The question of wealth generation vs redistribution is the single most important problem to solve in our social contract.
I have not seen any satisfying answer to it so far
Transmutation = alchemy?
Solves scarcity, but, to the point of your answer to the previous comments, there is a big, frightening, empty space...
Solving scarcity in land, materials and energy would generate oversupply of human capacity to work, or otherwise stated, of humans.
Can illimited intelligence integrate the need to solve wealth (if still a useful unit of measure) redistribution amongst "useless" humans? And can it do it without violence, forced elimination, and in a time framework corresponding to its advances in the other fields?
The three laws of robotics come to mind, and I am really interested in what comes out of the 6 months moratorium proposed by Musk & others.
Your post of today brings me back to one of your first non-covid publications. My fears were the same. I keep on reading all your publications with the same interest. Congratulations!
Glad to hear!
Based on what you say, you will enjoy this one:
https://unchartedterritories.tomaspueyo.com/p/future-humans-are-sporadic-indulgences
Yes, inequality is a very troubling problem today, in the Pre-Infinite Age. Well, let's be more precise: it's troubling to *some people.* It's certainly not something that the winners of capitalism are especially troubled by, beyond the crocodile tears from the Davos set. It's certainly not a dynamic that the people benefiting from are willing to change, from the wealthiest 1% on down to the merely comfortable "Dream Hoarders" in the upper-middle-class 20%. We could today feed the world with calories to spare right now, but we don't. We could redistribute wealth and income to satisfy at least a $10,0000 middle-income lifestyle for every one of the Earth's 8 billion people, but we don't. Seems an intolerable downgrade to those of us with more, even though it would fund a bounty that would make aristocrats of yore blush. There's really no upward limit on our "needs," when an American making six-figures dreams of becoming a millionaire and a millionaire dreams of billions. Being wealthy and high-status is more satisfying than being good and sharing, clearly. Don't you agree? Yes, YOU. Because you are certainly part of that group of relatively wealthy Dream Hoarders, aren't you? I am. And my progressive ethical or political beliefs, notwithstanding, when it comes down to it, I'm complacent to live in an unequal society. I'm happy enough to pay a hefty share of taxes on my income to fund transfers to others and to fund universal public services, sure ...but I'm certainly not giving everything I don't strictly need away to people who need it more. So, I'm part of the problem. We can justify it all day long, but we're just being unethically selfish by not sharing more in a world full of want. Not that you can really blame us for just being selfish, like most everyone else. David Graeber and David Wengrow's "The Dawn of Everything" thesis notwithstanding, it seems that inequality and elite efforts to maintain it are about the most consistent dynamic in the history of human civilization. Plus, exposure to wealth and high-status is proven to actually reduce our capacity for empathy, so, if "sharing is caring," then we, the relatively wealthy ones, have become uncaring in our comfort.
So, if you haven't gotten any satisfying answers to inequality, the (unsatisfying) answer is that wealthy and powerful people enjoy having more and that any solutions to redistribution feel like losing when you're on top. "Yes, but that's a malady of the Old World of scarcity," you're saying, "There cannot be inequality or want in the Infinite Age following the Singularity!" Infinity will render us ethical...
But WHY? What's the mechanism for change? Infinite intelligence can be a paradigm shift, but it is also force-multiplier or accelerant for existing conditions. Previous technological revolutions have tended to merely crystalize historical contingencies. It's a pretty arbitrary historical condition that the primary agents and beneficiaries of the computing age were white American men who had the means to study at places like Stanford (set up by a rather nutty wealthy heiress with railroad money... showing how the gains of previous technological bonanzas are laundered into the present). Were these West Coast dudes the smartest minds on the planet from the 1970s? Or even the best and the brightest from within the entire United States. They probably think so, and a lot of their fans agree, but the premise is absurd. They were certainly very clever young men, all... but standing on the shoulders of the existing power structure. And the other clever young men and women everywhere else weren't.
Conversely, historical dynamics don't let go of the people with less wealth and power, even after we have cool iPhones and ChatGPT. Americans argue to death why their Black citizens tend to have such stubbornly inferior lifestyle outcomes compared to white and Asian Americans. But isn't the answer as simple as: The former group endured 400 years of slavery, followed by another century-plus of explicit, legal, violent discrimination, and since then a few generations of more implicit structural racism. If that's not the reason, then it's a startling coincidence. This "natural experiment" in wildly divergent outcomes within the same society becomes even more clear when you look to the less remarked upon but even more terrible life outcomes of Native Americans, who never recovered from the historical sin of colonialism and genocide. No innovation dreamed up by Americans has changed this initial condition. Maybe because we can't. Maybe because we just don't care enough to try. And, though intentional advances against other aspects of inequality have been more successful, you still see the unmistakable mark of social conditions that existed four centuries ago in our country, despite all the creative destruction since. That's how path dependent inequality is. And it's not just the case in the US. In Italy, you can trace the wealth of the same wealthiest Florentine families all the way back to the Medici Era. In Sweden, that paragon of equality, the same eight families have dominated commerce and politics for centuries, in most cases back to Sweden's 16th Century independence and earlier!
So, why should an era affected by AI be any different? Our experience with AI certainly hasn't been encouraging so far. Building these models is the same group of American men (American IT workers are actually even more male than they were in the 1980s!), now joined by the imported elites of India and China. We've already seen that their models reproduce the attitudes and biases of the current power structure. AI still proves itself to be embarrassingly racist and sexist, which isn't at all surprising considering that it learns from existing human patterns. So, what mechanism do you imagine would make a more developed and capable AI something that would serve the interests of the many, rather than the few? After all, follow the money. Who owns AI today? What is it for? *Whom is it for?* Even the deceptively named "OpenAI" is owned by... Microsoft. It was a startup. But not a scrappy one. It was created by the same "PayPal Mafia" gang of Elon Musk, Peter Theil, et al. The ghosts of Internet 1.0 emerging again. And productized so far to drive Microsoft's legacy ad-driven search and monopolistic productivity suite businesses (and undoubtedly some national defense applications). The only other viable AI competitors today are from the same FAANGs who have gathered up all the treasures of the Internet 2.0 Era. As they have for two decades, these few companies deepen their "moats" by buying up any upstart rivals. Not that there's much danger of that. It's near-impossible to create an AI in your garage, anyway: AI requires tremendous resources--in data, computing, and training--to be viable. Network effects reign. Regular people will never be able to compete in this arms race. And, so far, no (democratically-controlled) state is stepping in to decentralize control, either. Even in "Communist" China, AI is conceived as a tool for building the national wealth and power. Not for making individual Chinese peoples' lives better. The Means of (Infinite) Production will everywhere be even more firmly in elite hands.
Just as with AI's precursors and antecedents--telecommunications and computing--AI will inevitably centralize wealth and power. In the same way as the original Gilded Age was made possible by how the Second Industrial Revolution undermined skilled labor interests, computing technologies and automation have made inequality *more* viable in this, our New Gilded Age. You just don't need as much labor, skilled or otherwise, to make a marginal unit of GDP today. So, you now get a society split into the 20% and the 80%. And that will seem tame, equitable, and "inefficient" compared to the Third Gilded Age of the Mid 21st Century. When the AI Wizards are conducting transmutation alchemy with limitless fusion energy, who needs labor or input from the poors at all! Or even the (formerly) normie middle classes, tapping away in their "bullshit jobs?" Just because we are freed from resource scarcity doesn't mean that there's any need to actually share those infinite resources. It will be the gods quibbling on Mount Olympus and the rest of us underfoot.
Are you willing to live on $10K/year?
If not, then scarcity definitely has an effect on generosity.
I think the calculations do change when we can have everyone get $1mm/year (in real terms) without effort.
The poor are not getting poorer, extreme poverty went from over 80% of the world’s population in 1800 to under 20%.
https://ourworldindata.org/grapher/share-of-population-living-in-extreme-poverty-cost-of-basic-needs?country=~OWID_WRL
Would be great if we actually see quantification of that claim / slogan about rich getting richer but mostly about the poor getting poorer. Last I checked China lifted 800m out of poverty in a single generation.
This is due to globalization, not automation.
Very interesting essay. But one quibble which is actually pretty major for undermining (pun!) your thesis: raw materials are (and will be) still scarce, even with the input of all the intelligence imaginable.
You said, "This is why peak oil has been declared for decades, but in fact we might well find peak oil not because we can’t find it anymore, but because we don’t need it anymore." Firstly, we have arguably already hit Peak Oil, in terms of peaking oil production. The IEA called for this year already. And, yes, that prognosis has been a slippery one, but we have some accumulating facts that point in the same direction: Conventional oil *discoveries* peaked in the 1960s. Conventional oil *production* peaked more recently, in the 2010s. Per capita, production's been going down since 1979, as population growth outstripped production increases. As for "proven" oil reserves... well, that's a very contested matter involving a lot of corporate and national secrecy and perverse incentives to overstate things. But, in any case, we're not actually pulling more oil out of those proven reserves, however large or small they may be.
Does that mean the oil ran out? No. There's still oil being produced, obviously. And there will be oil produced for some decades more, presumably. But the limiting factor has always been: Can a marginal unit of oil be can be extracted *economically* at a given price? If oil's $60/barrel, that reduces significantly the amount of proven reserves that would actually be worth tapping. And, at a certain point, unconventional sources are so energy-intensive, complex, and expensive to tap, they aren't reserves at all, anymore. Even at $120 or $200/barrel. Long after we stop producing oil, there will still be oil under the deepest parts of the ocean floor or aggregated with a much of non-useful gunk under the ground somewhere in Canada. But there it will remain, forever be too much trouble to extract, just like mining on an asteroid is likely to be.
And even if you take finances out of it, an even more limiting factor on whether an oil barrel goes from notional to actual is the Energy Return on Investment (EROI), or the energy input required to extract a marginal unit of energy output. And that's where innovation hasn't fixed our issue. Shale oil production, a significant innovation drawing from impressive amounts of human intelligence and skill, resulted in more oil production (especially in the United States), but the EROI decreased near to the point where it is costing nearly as much energy to get the stuff as it produces. After the initial rush of optimism and triumphalism, many of those unconventional producers quickly found that shale oil was never actually profitable, and was only feasible with a lot of easy financing. But it's not just shale oil running into this limit. Fracking shook loose a lot more methane, but at a cost. Getting the natural gas out of the ground and transporting it across oceans as LNG the way we're doing much more now since Russia invaded Ukraine makes that methane far more expensive. And, by extension, it makes energy too expensive to use, beyond a certain point. You see this with much of the German industrial base just closing up shop temporarily or permanently, given the expense of industrial heat from burning methane that now must be sourced as LNG, instead of piped from Russia. And, with oil, as we've already tapped 2/3 of discovered fields, see productivity of our most prolific fields trending in the wrong direction, and scrounge about for ever more marginal and harder-to-access discoveries, we'll be reaching the point where the EROI of fossil fuel energy production, overall, approaches 1. And that's very bad. One unit of energy to get one unit of energy is very different from the exponential effect that fossil fuels had for human civilization in the 19th and 20th Centuries, back when the ratio was closer to 100:1. A hundred years ago, you could be as inefficient with the stuff as you like because it was so cheap! Today, even with very mature fossil fuel technologies that approach the limits of efficiency, we can't get much more useful energy out of a barrel of oil that keeps getting scarcer.
But is this just because we're past the age of oil, and we "don't need it" anymore? Hardly! Oil demand fluctuates and there have been impressive gains made in replacing oil as an input for electricity production or passenger transport, but we're still very much in the oil age. Oil is still the only primary energy medium which is so energy-dense, stable, transportable, and versatile. We use fossil fuels for EVERYTHING. Including the manufacturing of the solar panels on my roof and the lithium-ion battery that helps my hybrid vehicle to go. We even basically eat the stuff, as you alluded to in your aside about fertilizers.
Oh, but we'll have a solution for that, too, you say. Though you didn't mention it under the Raw Materials section, perhaps nuclear fusion's limitless energy could allow for the transmutation of other feedstock into oil or other synthetic fuels? Perhaps we could create endless hydrogen? Perhaps, but this is relying on a kind of deus ex machina, isn't it? Viable fission doesn't even exist in prototype form yet, much less scaled out commercially over the whole face of the earth. And not for a lack of a half-century of trying. That transition, if it happens, won't be something for the first half of this century, if we see it this century at all. And aren't even the unimaginably complex machines that we hope will create fission manufactured with those same scarce inputs? You don't get a spherical tokamak fushion reactor without a hell of a lot of high-spec metals, specialized magnets, and high-voltage electricity. And you don't get those without the fossil fuel-driven processes that allow us to mine things, apply industrial heat to them, and manufacture and assemble complex machines to unbelievable exactitude.
So, I don't know quite techno-optimists think that nuclear fusion will be so easy and cost-free in the way that no other technology before it has been (remember that nuclear fission was also expected to be "too cheap to meter"). The costs in the inputs of financing and scarce resources will be largely front-ended, just like with nuclear fission, but they will be significant. And given the progress against this technology so far, we will be trying to build hundreds or thousands of fusion plants after 2050 just to keep up with our conventional energy needs, the time by which we're likely in the thick of a "polycrisis" of dwindling fossil fuels , essential resources (like phosphorus) that are also exhausted, energy that is expensive and unreliable, food production that is insufficient, all the various other knock-on effects of Climate Change, and all the political and geopolitical effects thereof.
Thank you, very thoughtful!
The entire thesis of today's article is very much that an infinitely intelligent AI is a Deus Ex Machina. It can solve problems in ways we can't even foresee.
That said, I'll give you a path *we can already foresee* for energy.
Drop photovoltaic costs enough (as it's clear it's happening) and energy soon becomes too cheap to meter. With that energy, you can generate all the fossil fuels you want just by extracting CO2 from the atmosphere, in a cheaper way than pumping them out of the Earth. Another one you can easily do is drop nuclear fission energy costs tremendously, because it can simplify the designs while improving them, in a way that they're much safer than any other energy source (already true today) and also cheaper (not true, but because of a limit in intelligence).
So infinite intelligence does solve the problem of not enough energy or fossil fuels.
The transmutation argument is different. It's not for the scarcity of molecules (simple to solve with infinite energy), but for the scarcity of elements (like Bismuth or something).
I don't think it's so hard to imagine infinite intelligence solving the issue that there are trillions of Suns already, but we just don't know how to make them and use them. But if you can help me understand what's the actual physical barrier to that, I'd love to hear!
It's a common working assumption that some form of AI now in development, including ML, LLMs, and other forms of artificial intelligences not yet conceived of--will at some point hit that "Singularity" and then beyond to an exponential age of intelligence that makes Moore's Law look trite.
Maybe.
But, again, AI has been in development about as long and gone through even more hype-cycles than nuclear fusion. Skepticism for both is warranted. Exponentiality tends to run out of steam. Especially when the initial conditions for progress don't continue indefinitely (more on that later).
But let's assume that we can and do supplant the Intelligent Design God of the 18th Century imaginings, rearranging the very molecules of the Universe in ways that are most useful to us. Still, even there, gods and other magnificent beings run into limits. The God Yahweh had limits. After six days of Creation, He had to rest, according to Genesis. And that was just the beginning of all the frustrations with His Creation! There's book(s) about it. Very popular reads. But, look, too, at the more empirical facts of our existence: the Sun itself has limits! It creates energy "for free" insofar as it self-immolates and will expire in 5 billion years or so (which isn't as relatively long-seeming in Sun terms). It will start to expand outward as it ages, like so many of us, losing its youthful heat and vigor in the process as it burns through initial endowments. Even the wider Universe has limits! As it ages, it also expands and cools (we think), which leads inexorably toward the Heat Death of the Universe (presumably). That's a bummer for any Demiurge behind it all because even information itself may not be able survive that (we think). The final cost of the Big Bang? All that doing smart stuff to just be forgotten!
I raise these extremities to make a point, but pull the bounds closer to our scale and you see that there are limits here, too. Everything we use now was created via incredibly powerful celestial or geological forces beyond both our agency and ken, burning through unfathomable amounts of the ultimate scarce and non-renewable resource: TIME! Given the evolutionary nature of the Universe, which has operated over a 13-14 billion years of that ultimate scarcity creating practically infinite variations of matter and activity, isn't it a bit hubristic to think that we can recreate the magnitude of that power without even as much time involved?
Could we use super-intelligence to turn a mountain into a freshwater lake? Well, the Earth did it all the time, but it took a few million years. Could we turn all the Crocs that more primitive humans created in the 2010s into something non-renewable like helium so that we can have funny voices and asthma sprays forever? Well, the Universe kinda basically did that, but it did take a few billions years. So could we accomplish the same in a few hours? With enough energy; from enough intelligence? Maybe. Energy is all around us: There are an estimated 47 TW of heat energy flowing from the Earth's core. Even more energy, 173,000 TW of solar energy hits the Earth from the Sun. The rest energy embedded in all the matter on Earth is incalculable. This is A LOT of juice! But, remember, that energy does something already. A lot of it is not available for human consumption. Or at least not without displacing all the other "ecosystem services" that exist outside of our direct interest, which we have done in so many other aspects of life (e.g. displacing almost all large terrestrial mammals, to make room for our meat or blowing through nearly all the arable land on the planet for our cereals).
So, even if it's possible to capture and apply all this energy, what's the equal-and-opposite reaction? What's the unforeseen externality? What's the cost? If we haven't been intelligent enough to look out in front of our own cleverness to foresee the consequences of our own cleverness, will an AI? Why do we assume that an AI would be "wiser" and more long-termist than us, rather than just as monomaniacal and instrumental in thinking as we have been? Carpet the entire planet with super-efficient solar panels and drill boreholes to suck up the last residual heat we can get from the Earth, in order to put on the real show of rearranging the entire Solar System for post-scarcity profit! Because there is never "enough." If there were, why would we have billionaires today?
This post-scarcity bonanza could also inevitably result in wars on an unimaginable scale. Because your summation for the reasons for war (and their obsolescence) could be naïve. Any good Space Opera will remind you that "scarcity" is always relative and more infinity is always better than less. Even the utopian society of Star Trek has fighting. So maybe we'd (or rather the AI versions of us) would just be wringing the last available units of energy from our Earth and its Sun to power weapons. Weapons that would become their own justification as adversary intelligences also matched them with weapons of equal and greater potency. A Universe-wide arms race hoovering up all the matter and energy until... well, something terrible that fast-forwards the timeline to the Heat Death of the Universe for the expiration of the intelligences that the Universe sprang forward to know itself.
But, again, that's *if* we (or the aliens) could even pull any of this off. It's a good bet that we also just will never be able to. Or that an intelligent species like ours could, in theory, but always ends up destroying themselves with other means long before (Fermi Paradox). Circling back to my point about the timeline of fusion: What if we *could* make it happen, but won't ever quite get there because our other scarcities catch up with us first? We hit a wall from 2030-2050, where there's too much instability and increasing scarcity in the necessary resources to finish the work of commercializing and scaling out fusion, AI, or other exponential technologies? Could we find ourselves like a race of Nikola Teslas, full of genius ideas never quite fully realized in a too-short life ending in frustrations and penury?
If we sum up what you say, there are two limits to scarcity:
1. Physics
2. Non-physical stuff, like for example things tied to humans (love, sex with a human, social hierarchy)
We agree on that.
But I think physical limits are extremely far from anything imaginable today, so compared to the world today, we will reach many physical infinities. The same way as today heat, electricity, water, and computing power are near infinite for you, so can most other physical limits be after intelligence becomes infinite.
Then there's point number 2, which I explore in the article this Thursday
Maybe we should invent AW before we invent AI?
Great piece. Really laid it out clearly
I started a company last year that uses data analytics (our AI still being built) to make land use and human work more efficient for the deployment of renewables (won’t name it as I am anon here). Your outline is a more all encompassing description of what we talk about in some of our longer product meetings, cool article!
Glad to hear!
and creates 1 big one: Skynet!
This is absolutely brilliant. Such a great post.😀
Glad to hear!
Re : scarcity of ressources. Don’t forget asteroid mining. With fusion propulsion it shouldn’t be a problem. Will bring a *massive* amount of raw material
You will love my next space article! It will probably come in the next few weeks, when Starship launches