An Economic Singularity

Here's a pathway to a sudden and catastrophic economic Singularity that doesn't require AGI.

Let’s take a break and think big picture.

I was first introduced to the concept of a Technological Singularity (“The Singularity”) in 1998, when I read Vernor Vinge's brilliant essay. It’s worth a read if you haven’t read it before. In short, it argues that the pace of technological advancement has been accelerating, and that once AIs emerge, technological advancement will shift from linear to exponential growth, changing everything seemingly overnight (think of it as crossing the event horizon surrounding a black hole; once you cross it, the rules of physics break down and anyone crossing it will be spaghettified as the parts closest to the singularity at the center acclerate towards it faster than those parts farthest away from it).

Fortunately, the potential for a Technological Singularity (where AI becomes smart enough to improve itself at ever-increasing rates) isn’t likely to occur in the short to medium term.

• The rate of improvement seen in AI models (advancing towards AGI) is slowing (diminishing returns) despite massive investment. NOTE: AIs currently ‘write’ 20% of the code used to design the next generation of models.

• This rate of decline in improvement has forced the leading companies to focus on mass market adoption (consumer AI companions and agents, aka Social AIs).

• Efforts to build persistent and coherent Social AIs (running on top of commercial LLM platforms) that don’t suffer from hallucinations and entropic decay have hit a wall as well (usability limited to hours of use rather than days, months, and years).

However, while the potential of a near-term Technological Singularity is diminishing, it’s still possible that we’ll see catastrophic singularity events in discrete areas, such as economics. A good example of this is a recent paper from a group of economists from MIT and Harvard (“The Coasean Singularity”). It argued that;

• Firms form to minimize transaction costs (Coase).

• AI agents will significantly reduce transaction costs across the entire economy.

• If transaction costs decline markedly, the need to create firms would evaporate, necessitating a complete reevaluation of economics and market design as existing corporations dissolve and new markets emerge.

While a Coasean Singularity would be disruptive, it would be a minor event compared to a true Economic Singularity, which completely reallocates the production and allocation of wealth faster than we could respond. To fully explore this, allowing us to identify something similar in the future and react to it, let’s map a pathway to a sudden and catastrophic Economic Singularity.