A good few years before the AI craze, my Oxford lecturer gave a presentation on the shifting nature of work. An economic historian by trade, Judy Stephenson traced the arc of compensation from labor market considerations in early modern London, and wove a full-circle tale of being paid piecemeal (output) in the nineteenth century, to hours (input) for most of the twentieth century, and then back again in the twenty-first-century gig economy.
The delivery and ride-sharing services were the major concerns of the intelligentsia during the 2010s. You were paid not for your time but for the output you quite often physically delivered, with resulting debates over unions, safety, and wages.
Stephenson accounted for the changes on very Coasean terms: In the assembly-line work of a century ago, it wasn’t worth the transaction costs of figuring out exactly whose contribution was worth how much, so you just roughly averaged out everyone’s hours with some extra perks for responsibility or long service. And compared to the at-home weavers of the century before, it was much less clear who was responsible for the exact value-add. Put differently, the loss of efficiency associated with time-based pay (free-riding, monitoring, slacking off, or shirking work) might have been less than the costly efforts necessary to constantly re-establish rates for specific tasks.
Economic textbooks, heavy on the modeling, might imply that performance pay is more efficient since it aligns incentives and minimizes free-riding. Enter computers and digital markets matching supply and demand, plus standalone gig workers entirely responsible for their own output. Those institutional changes shifted the bargaining power and the Coasean transaction costs involved — making the real world so much more like the sketched model of an economics textbook.
Easily Replicated Abundance Meets the Economics of Infinite Content
There’s an obvious self-selection in the current labor-related worries coming our way: It is precisely those of us who have invested most in this credentialist commentariat, who have sacrificed our lives and oriented our identities around the very cognitive and generative skills that LLMs now so effortlessly replicate.
It’s no longer that hard to have ChatGPT write like me (just train it on my past writing), Claude to code like a programmer with a decade of experience, or have a combined AI effort produce a beautiful, two-minute, period-piece ad spot for $100.
In The Great Harvest, a recent and ironically mostly AI-generated book, Adam Livingston captures the white-collar workplace revolution underway: It’s “not that your career will vanish overnight but that it was always just a fragile assemblage of solvable problems, [… your job] was actually a collection of separate functions waiting to be identified, isolated, and optimized away.”
The music industry and the economic value of songs were early indicators here, with supply and production way surpassing any feasible consumption or customer on the other side. Even though its economic threat stemmed originally from pirated rather than easily replicated material, the marginal value unavoidably fell to around zero. While Taylor Swift rakes in royalties from streams and other artificially scarce legal arrangements, she generates more economic value from concerts and merch. Her physical being becomes the ultimate, rivalrous, nonreplicable luxury good.
With the marginal cost of producing videos, images, music, or words going to zero, we should have expected infinite content and next-to-no meaning — see YouTube, TikTok, or Twitter.
With the rest of the arts and white-collar knowledge industry up next, it’s a little bit of an economist’s puzzle why prices (i.e., wages) haven’t dramatically fallen yet, to adjust relative scarcity and the now much more abundant supply — stories of social anchoring or nominal contract rigidities, no doubt. So far, we’re much more likely to see quantity adjusting, meaning fewer workers or worse labor-market conditions for programmers, journalists, accountants, and other white-collar jobs.
Where’s the Value? Humans as Tastemakers
A lot of digital ink has been spilled on trying to identify where we go from here. In a world of informational abundance and adequately generated text at everyone’s fingertips, where’s the economic value?
“Brainpower is now a commodity that is going cheap,” Andrew Yang reflected this month. Perhaps the best thing we can say about his UBI-infused presidential bid in 2020 is that it was premature.
“We have a love-hate relationship with working for a living,” Tim Harford observed for the Financial Times; the pain and hardship of working is heavily bound up with meaning and identity. Fred Krueger and Ben Sigman, in another recent book, observe that the “labor theory of value collapses when machines do all the labor,” and that “scarcity pricing becomes meaningless when AI makes many things abundant.” As intelligence becomes infinite, they conclude, the finite becomes infinitely valuable.
These reflections might as easily have been titled “The Return of the Labor Theory of Value,” not because the LTV was a particularly revolutionary economic theory, but because of what it indicates about our infinitely replicable information and knowledge system going forward. If everything from music to code, words, and video can be created at the press of a button, the only scarce thing left beside the physical world is our human attention. The things we choose to do, choose to look at, choose to labor on.
Fiction writers, faced with the nearly infinite onslaught of storylines and millions of predominantly self-published titles each year, have realized this: Their words or imagined characters may not be scarce, but the very fact that they labored intensively over them is what other humans recognize as worthwhile and impressive. (We might ultimately decide to pay a premium for human connection, attention, or presence.)
Book sales, while pretty stagnant in nominal and real terms, might be monetary votes of appreciation more than actual desire or follow-through to consume the work.
In the past, these industries had an overhang of gatekeepers and tastemakers deciding what was good music, good art, good writing, or good journalism. In the past few decades, it might have been liberating to have the gatekeepers shoved aside via technological means, but it’s only now that they’re gone that we’re starting to miss them. The artificial scarcity they imposed elevated excess economic value onto songs and books and movies that can now be generated and duplicated by the millions.
One way out, then, is to recreate the gatekeeping — not in production, that ship has sailed, but in attention and awareness. We might look to respectable minds, like we once did respectable labels or studios or outlets, not for reporting what is in the journalists’ style, but what matters. Trusting in their vision of what matters, using their long and somewhat obsolete experience as a filtering mechanism against the information overload we’re otherwise doomed to.
Muscle lost its economic dominance long ago; we all know that story. Now that machines are coming for the brains, we have a similar story of scarcity, abundance, and obsolete skills to contend with. What remains scarce — attention, trust, physicality, judgment, embodied presence — will command the rent.
