Thought Shrapnel

Note: as regular readers will be aware, my habit is to quote part of the excerpt as the title of Thought Shrapnel posts. I, personally, am not advocating for abyss-directed activity.

I’m not sure who the cipher “aethn” behind this essay is, but this is a pretty standard argument dressed up in fancy (and somewhat eschatological) language. I’ve tried to excerpt the main thrust, which is: LLMs are getting better and seem to be starting to replace some lower level jobs; this will continue and cause a rupture in the fabric of society. Somehow we need to prepare for this.

While I do believe that AI is somewhat qualitatively different from previous technological inventions, I’m also a student of history and so know that disruption doesn’t happen everywhere all at once. As William Gibson is famously quoted as saying, “The future is already here — it’s just not very evenly distributed.” Just because some people, like the author (and like me), are messing around with LLMs and finding them powerful for our work, doesn’t mean that everyone else is.

Essays like this tend to miss out existing inequality in a rush to talk about future inequalities. And, it has to be said that our western economic and political structures have proven remarkably resilient to a number of shocks over the past centuries. Fredric Jameson noted that, “it is easier to imagine the end of the world than to imagine the end of capitalism.” I’d note that, sometimes, one person’s “violent revolution” is another person’s evolution of capitalism into a new form.

We are at the precipice of a revolution more violent than the Industrial Revolution. This revolution is not about the typical vulgar parochial anxieties on job security—although it is part of it—it’s about a violent upheaval of the very socio-economic fabric in the way our world is organized.

[…]

The latest innovations go far beyond logarithmic gains: there is now GPT-based software which replaces much of the work of CAD Designers, Illustrators, Video Editors, Electrical Engineers, Software Engineers, Financial Analysts, and Radiologists, to name a few. This radical automation exists without any sophisticated fine tuning or training.

[…]

The frequent naive view of the amount paid in a wage is that it’s proportional to the difficulty of the job. If this were the case then certainly all wages will substantially be decreased with the advent of LLMs and that regular economic structure will be maintained. Instead, through the normal polemics we find that’s not the case.

Wage is instead best viewed from the perspective of the profit maximizing economic agent, as in what motivates such an agent to accept a wage from another party at all. If such an agent were able to endeavor alone and capture all of the value from its enterprise it would do so. […] [T]he agent must determine if an offered wage is greater than the expected value of its solo enterprise. We then find that wage must be greater than the expected value of the opportunity cost of the uncaptured labor value incurred due to employment. For much knowledge based work, this is acute since with the same skills needed for employment one can make a competing enterprise to their employer and capture all the value. Other professions require you to have large capital to do so, so the opportunity cost is either non-existent if you cannot access that capital or further discounted by the financing cost and the risk.

[…]

Generally aligning a team on a common product requires expensive payments to each member to provide contributions. Indeed, the early-stage venture capital industry relies on this fact.

The latest LLMs make such a provision completely redundant. The proprietor of today can supervise and delegate tasks required to build a product to the latest GPTs in virtually any knowledge field. The single proprietor now has the efficacy of maybe a team of 6-10 conservatively and further is able to produce even higher quality work.

[…]

One may demur and claim there simply won’t be enough knowledge-based services in demand, however we instead find a Jevon’s paradox, where demand for these services increases. The reduction in costs of producing the same services will make them more ubiquitous and bespoke at even a per individual basis.

[…]

Corporations whose products have no moat by economic scale or network will be forced into specializing their products as they surrender market share to the sea of companies. In the Deleuzian fashion, proprietors can build almost as quickly as they can imagine, as predicted, rattling the foundations of the economic order itself.

[…]

The social order must change drastically to a future where institutions are no longer designed to feed corporations future employees, they merely won’t harbor that demand. Many existing knowledge based workers will largely have no choice but to engage in enterprise. Educational systems must adapt and accommodate this new entrepreneurial exigency for labor in the new economic order. Workers of the future must be emboldened to eschew wages in favor of dropping into the abyss in order to have any meaningful income.

Source: aethn’s essays

Image: Maksym Mazur