At no point in recent history has the future seemed so uncertain. All across America, in cavernous data centers with blank facades, a new order of AIs is taking form. We are told that in the coming years, they may usher in an age of abundance, wipe us clean off the Earth, or something in between. You may wonder whether it will soon mean something different to be human. And you may have more practical questions, too—about your personal finances, for instance. On Wall Street, the quants are already sketching out the likely economic scenarios, but normal people also want to know: What should they be doing with their money, so long as they (and their money) continue to exist?
Imagine the rosiest AI-abundance scenario. The “compressed 21st century” of scientific research prophesied by Anthropic CEO Dario Amodei actually materializes. Instead of killing us, the intelligent machines cure our cancers. Their superhuman productivity drives the price of most goods and services to near zero. We find ourselves encased in cocoons of unimaginable luxury. I once asked Sam Altman, OpenAI’s CEO, what this future might look like. He told me that solar-powered robots would build us any house of our design for a trivial cost. No one would go unsheltered, let alone hungry. By today’s standards, everyone would be very rich.
Choosing the best financial strategy in a world like that won’t be an existential concern, but there will still be different levels of wealth. People might vie for what the economist Fred Hirsch called positional goods—scarce stuff whose value depends on other people not having it. Some research suggests that housing ranks highest in the hierarchy of these goods. Even if, as Altman told me, construction will be a solved problem, location is everything. There are only so many buildable lots along California’s Santa Barbara coast or in Manhattan’s West Village. Get one now, if you can, and stock up on collectibles that are verifiably rare. If you’re already rich, buy that Fabergé egg and that T. rex skeleton at Sotheby’s. And if you aren’t rich, at least save your baseball cards.
Of course, we may not get an AI utopia. Molly Kinder, a former Brookings Institution fellow, has described a “messy middle” scenario where a stretch of profound job disruption and economic chaos precedes a more settled AI future. Even comfortable white-collar workers could encounter a gantlet of financial dangers. The first thing you might notice is that you’re no longer getting raises. Your boss—and their boss—may respond to the historic level of economic uncertainty by pushing back their retirement. They may stop hiring entry-level workers. Your corporate ladder may shrink down to a few rungs. You’ll spend more and more of your workday reviewing AI output. On your way home at night, you’ll wonder when machine intelligence will be able to do your monitoring work, too.
In this scenario, you’ll still have a mortgage, tuition bills, and perhaps elderly parents to care for. You won’t want to tap your retirement accounts to make up for any shortfalls, because your middle years are prime compounding time for your 401(k) and Roth IRA. And if you claim Social Security early, you’re locked in at a lower benefit—a dicey move if AI eventually ends up extending your life, as the AI kingpins have promised. So you may just have to hold on, keeping your expenses low and being ready to relocate on short notice, anything to keep hopping from one employment life raft to another before obsolescence sets in.
That may sound grim, but things could be worse. Imagine the scenario in which AI triggers an unprecedented wealth-creation event, but only for the relative few who have significant equity in the companies that created it. Everyone else might be relegated to a permanent underclass, as some in Silicon Valley have come to fear. If you think this will be our future, try to join the elect, if not by marriage, then by shifting as much money as possible into shares in the leading frontier labs. But that may be difficult, because at least for now, OpenAI and Anthropic are still private companies. Buying their stock on the secondary market is not easy. Equity stakes in these companies have become so desirable, in fact, that one home seller in Sonoma County, California, offered a $500,000 discount to buyers who could pay in Anthropic shares.
You’ll also need to make sure you pick the right company in which to invest, and purchase a large enough stake for it to matter. If a superintelligence learns to improve itself rapidly, the race among the frontier labs might be winner-takes-all, and they currently jostle for position from week to week. Timing presents another challenge. Leopold Aschenbrenner, the former OpenAI researcher who founded the hedge fund Situational Awareness, decided to bet on companies that are likely to benefit no matter which lab wins. He may have gone in too strong and too early. His fund was up more than 400 percent earlier this year, but then during a market dip, his public-equity portfolio quickly lost more than two-thirds of its value, and he was forced to sell the bulk of it in a single trade.
Even if you pick the lab that creates the first superintelligent AI, and you’re one of the lucky few who gets unimaginably rich, you’ll still be subject to political risk. If a lot of people suddenly find themselves reckoning with a lifetime of impoverishment, they might not be delighted by your success, and they won’t be entirely powerless. Provided that democracy survives, they’ll have the vote. They may make clear to their congressional representatives that they’ll accept nothing less than a profound redistribution of the AI windfall. They might pass a bill that makes the New Deal look like Reaganomics. (A 2020 paper by the Centre for the Governance of AI urges AI companies to pre-commit to redistribution to avoid this fate.) This bill might aggressively tax all AI revenue collected by the winning lab or nationalize it outright.
And what about the really bad scenarios, in which AI goes rogue, or is used for evil purposes? If automated agents give the pillars of human civilization a good shaking, that could imperil your net worth, along with everything else. Earlier this month, Altman said that embracing this technology means accepting that “some bad things are going to happen.” Greg Jensen, the managing chief investment officer at Bridgewater Associates, one of the world’s largest hedge funds, has said that the risk of a “major financial incident run by AI” or “physical disaster” in just the next few years is “way higher than anybody should be comfortable with.” Anthropic’s recent filing in advance of its IPO warned that such an incident could cause harm on a scale that no single insurer, and potentially not even the entire insurance industry, could cover.
[Read: The 24-year-old who lost billions]
Experts have long warned about the possibility of an AI-led cyberattack on financial institutions, perhaps one that scrambles bank balances, rendering all account records suspect. Some of these institutions have implemented backup systems to avoid a great reset of this kind. Banks may record their customers’ account data every night in a standardized, encrypted format, and then send it over to an air-gapped archive. They may also partner with a “restoration” institution that would allow depositors to see their balances and make withdrawals in case of an emergency.
But future AI agents could find some way to compromise this system too. Stuxnet crossed an air gap by way of thumb drives, and it was designed by human hackers more than a decade ago. There may be no truly satisfying hedge against this kind of attack beyond keeping paper records, spreading money across many banks and brokerages, and stashing emergency cash—in small denominations, of course—at home. Gold could be an option, too, and there is some evidence that Silicon Valley types are buying lots of it. But gold is difficult to store and prone to being stolen, and most people can’t make change against it.
[Read: What if AI is the end of the world?]
Compared with some of the catastrophes that AI luminaries warn us about, a global financial cyberattack would be quite mild. Amodei has said that a superintelligence could one day help a bioterrorist engineer a new pathogen, custom-made to trigger a pandemic—perhaps one that would be far more lethal than COVID. An AI used by the military could coordinate with others to gin up a nuclear exchange. There could be some other rupture that we cannot yet imagine. Many people would likely survive these catastrophes, but civilization wouldn’t carry on as usual. For a time, the world may descend into a more Hobbesian state, in which brokerage accounts—and other aspects of the formal economy—are distant memories.
For decades, preppers have been hedging against these more extreme possibilities. They’ve secured remote parcels of land with vegetable gardens and working wells. They’ve stockpiled guns and ammunition that can be traded in a pinch. Billionaires have done high-end prepping of their own. Some are acquiring vaultlike bunkers in New Zealand. (Altman has said that his emergency retreat is in Big Sur.) These backup plans have failure points, too. Transportation to a bunker could be difficult, depending on how bad things get on the way to the airport or in the sky. Unfriendly locals may be waiting for you on arrival. Even if you get down into the bunker, your armed security may decide that, actually, the future should belong to them, not you. Some researchers have suggested that trying to wait out a catastrophe in pampered isolation may be foolish. They say that living in a close-knit, resilient human community may be the best hedge of all.
There is one extreme scenario that no one needs to bother prepping for. Unfortunately it’s the worst one: extinction. You can’t hedge against a total apocalypse, except by buying indulgences. If you think AI will usher in doomsday, spending down your nest egg might be the most rational thing you can do. Even now, there are people who have convinced themselves that because AI will kill us all, they have no future to plan for. Nate Soares, the president of the Machine Intelligence Research Institute, told my colleague Matteo Wong that he’s stopped saving for retirement. “I just don’t expect the world to be around,” he said last year. But to be clear, that’s not financial advice; none of the foregoing is, and least of all that.
Leave a Reply