Agentic
Everything is amazing, and everyone is furious.
Everything is terrible, is the loose consensus across America, Canada, and the UK, each of which I know well and have lived in for multiple spans across many years. Deregulation is empowering corporations and the rich to cruelly exploit the poor and middle class; or, alternatively, excessive regulation is choking any hope of real growth. I say ‘middle class’ but there isn’t even a middle class any more, just a precariat who will never be able to buy real estate. Climate disasters will doom the poor South, while here in the wealthy North, the (conservatives|woke), who are all in their secret evil hearts oppressively (fascist|communist), are winning the culture war1.
And yet. At the macro level, by many measures, things have never been better! Deaths from air pollution, heart conditions, and cancer, the leading causes, have plunged. Life expectancies have risen globally, and while the one-two punch of COVID pandemic and opioids epidemic famously caused US life expectancy to drop, despite COVID and opioids, it dropped only to the level of … 2003. Median real income has increased in Canada and the US, and … well, it hasn’t fallen in the UK. Climate change actually looks solvable, and per capita CO2 emissions have flatlined globally while plummeting in the North.
Meanwhile, science and technology have ushered in:
An era of clean energy superabundance.
An era of space exploration at a scale previously limited to sci-fi pipe dreams.
An era of miraculous vaccines, which in turn are only the tip of the biotech iceberg; scientists—plural—are suddenly curing their own cancers.
All at the same time! Heck, we even found a cure for obesity. Oh, yeah, and almost every human being carries with them, as a matter of course, a supercomputer with instant access to ~all human knowledge.
And yet: everything is terrible!
…What is going on?
The siphon and the forge
There are basically two ways to gain wealth, both individually and collectively. One, the forge, is to make and sell something new: a new business, new factory, new technology, new service, or new insight. Call this the forge approach. The other is to find an existing flow of money and divert some of it to you; the siphon.
When you read about private equity destroying well-known companies, or Cory Doctorow-esque “enshittification” of successful online services, or the “resource curse,” or Wall Street firms slicing off extra basis points courtesy of high-frequency trading, or big-box stores strong-arming providers into the creation of retail deserts, you’re reading about siphons. When you read about Nvidia or SpaceX or mRNA vaccines or Terraform, you’re reading about forges. Blockchain initiatives are supposed to be forges but almost always turn into siphons. Boeing and GE and NASA infamously morphed from forges into unholy chimeras of forge and siphon. Building housing is forge work; NIMBYism is siphonage.
There’s nothing intrinsically wrong with siphons, just as there’s nothing intrinsically wrong with governments of oil-rich nations claiming a royalty on their exports. But in practice, most of the time (though not in Norway!) this turns into the resource curse; corruption creeps in, a small minority enrich themselves, the smart and ambitious and educated focus on getting a bit of the existing siphon rather than building a new forge, this status quo is stabilized and perpetuated by the money fountain … and possibilities stagnate for everyone who doesn’t manage to suckle on the siphon.
The fundamental difference, and the fundamental problem, is that forges are positive-sum, increasing agency for everyone, while siphons are zero-sum, doing so only for a lucky few, and decreasing it for the rest.
Agency
Years ago I worked with two Romanian expatriates who had fled their oppressive homeland. At first the only place they could get to was South Africa, at the time so crime-ridden that cars were equipped with flamethrowers to drive off assailants. Eventually they made their way to Toronto. They were strangers in a strange land, with no safety net, no support network, no family to back them up…
…and they seemed enormously happier than many of the wealthy, well-educated scions of upper-middle-class Canada that I knew, because they were the unquestioned maverick captains of their own lives. They knew beyond doubt they were capable of doing extraordinary things to reshape their lives again if / as / when they needed to. They had agency in spades.
The ‘paradox’ of people thinking everything is terrible, while so much is so much better on the macro level, is no paradox at all. Day-to-day life has gotten enormously better — medical care, technology, the overall production values of everything from coffee to cars — but at the same time the average person has lost agency over the last ten years. We have built a better world in which people have less control over their lives. Their cars are much improved but they have less opportunity to go somewhere great.
This is most obvious with housing. At the turn of the millennium, an ordinary person could move to one of their continent’s great cities — NYC / LA / SF in America, London / Paris in Europe — and live there on the proceeds of an ordinary retail or clerical job. It wasn’t easy; you’d probably have to live in a hostel for the first month, then with dubious roommates, and worry often about money; but you could do it. (Admittedly partly thanks to the generation-long hangover from the 70s, when people fled the “danger, decay, and paranoia” of those same cities) It was an option available to you. You had that agency.
Now? Not so much. Landlords are the enemy and real estate, of course, is completely unaffordable. …Well. …That is to say. Housing in places where everybody wants to live is unaffordable. In Bradenton, Florida, a nice little city a short drive from beaches & Tampa & Sarasota, you can buy a 1BR condo for $95K and a 2BR for $200K. In fact there are plenty of affordable beachfront towns in America. You can live there on an ordinary job. …But you can no longer live in San Diego, or Venice, or Miami. This better world has closed off that option.
The desire for agency has turned university from a relatively languid, semi-optional period of post-adolescence into a desperate Darwinian battle to get one’s kids into the Ivy League, in the hopes that they will have only a little less agency than their parents once enjoyed as birthright. Which in turn means endless tutelage and extracurriculars and volunteering and even less agency for aspirational teenagers. And for those who don’t shoulder their way onto the success track, the sinking feeling that they have no agency at all, that they will be buffeted by the winds of economic squalls and vicissitudes their whole life.
Over the last 25 years Silicon Valley brought us enormously more agency: Amazon, Google, Uber, Facebook, WhatsApp, Twitter, all broadened the possibility space of everybody who used them. If we stipulate that they have since been enshittified (I think that’s only true for some of them…) then the problem is not that they now suck. The problem is that there’s nowhere else to go to gain the same agency — say, the ability to keep up with your friends’ lives, and what they found interesting, on an ongoing basis, without being constantly oppressed by unpleasant experiences while doing so — they once bestowed on us.
Climate solutions? Space exploration? Better food and coffee and beer and cars and clothes and televisions? 1,000 new streaming series for $13.99 a month? None of these do aothing for the ordinary person’s ability to command the course of their life. Medical advances? They forestall and prevent sudden and drastic losses of agency … but that’s not the same thing. As for artificial intelligence — well —
AIgentic
You might have expected a piece with this title on this Substack to be about AI agents … and if so, you weren’t entirely wrong. I think you can see where I’m going. Modern AI which increases the agency of ordinary people will be welcomed. Modern AI which does not — which is used as a siphon — will be seen as another unwelcome, invading force.
Right now it is … both? Modern AI lets people craft words, images, sounds and videos that they previously could not (to the great dismay of most of those who previously did so for a living.) In that sense it immensely increases the ordinary person’s agency. Furthermore, many millions of people already use ChatGPT or Claude as an ongoing amenuensis, and many millions more would if they realized how much it can help them make sense of the world, themselves, and their tasks/problems. (Yes yes in an imperfect sometimes-wrong way, not unlike a human amanuensis.)
But. On the other hand. As more and more processes are automated which couldn’t have been before, because they require judgement or fuzzy logic or context that programmatic software could not manage, people may feel that they’re losing collective agency, that humans are no longer in charge, that they can no longer go to an actual person for assistance with, or repair of, a system—and they may be right.
(This still might be preferable to the “phone tree from which you cannot reach an actual human” corporate antipattern; there’s a darkly funny outcome in which people of the 2030s think to themselves “welp, our AI future is pretty bad … but not nearly as bad as the early twenties were!”)
(Plus of course the people who themselves are automated will definitely not feel like AI has increased their agency, until and unless they find a new and better job.)
We’re seeing this already online; we can all agree that ‘AI slop’ is often no worse than the ‘human slop’ one often has to wade through to get to the decent stuff on Twitter / Reddit / YouTube / etc., …but we, and I very much include myself, still have a vastly greater aversion to the former. Any AI products which feel like they’re getting in the way of what people want will fail; I fear we’re going to see a lot of e.g. AI travel agents which keep offering almost-right or not-quite-right options, until the user thinks to themselves in frustration “I should have just used Expedia” and gives up.
…But we’ll also see ones which leave the user calling out, wide-eyed, “Honey! This says we can spend the long weekend at a four-star hotel in Cancun for five hundred dollars per person round-trip!” At the individual / retail level, the market will take care of agency through AI. It’s at the systematic or institutional level where designers and directors and engineers will have to ask themselves: “Will this increase people’s agency? Or … or just ours?” And, more importantly, will have to care about the answer.
Stipulated that there are real fascists and real tankies out there, and the former are a much more appalling problem because they are today much, much closer to the levers of power. But both sides of the culture war claim—often (though definitely not always) in good faith—the other side is inherently authoritarian and oppressive.