AI – Chasing Tales

This is not a regular post, just me musing out aloud here. AI is economically disruptive not because it is intelligent, but because it behaves unlike anything our existing factors of production were designed to describe.

Economics does not have a formal checklist for what qualifies as a factor of production, but there is a recognisable pattern. A factor tends to be:¹ ² ³

A necessary input to production (you can’t produce at scale without some of it)
Scarce relative to demand (so it commands a price and has an opportunity cost)
Distinct enough that tracking its quantity and return separately actually improves our understanding of the economy

This is how we ended up with land, labour, capital, and entrepreneurship.

FoPs also have their own characteristic of return:⁴

S. No.	Factor	Return
1.	Land	Rent
2.	Labour	Wages
3.	Capital	Interest
4.	Entrepreneurship	Profit
5.	Artificial Intelligence (?)	Data/ Information (?)

What stands out immediately is that all traditional returns are monetary, because economics measures factor rewards in money terms. A person lifting a bag and moving it somewhere else is not “producing money”; they are supplying labour that is then valued in money. At the moment we don’t have anything like a standardised, broad market that prices raw data or AI outputs in the same way. AI primarily produces streams of information—predictions, classifications, strategies, compressed knowledge. Money appears later, once those outputs are embedded into decisions and products.

Another difference is agency. All existing factors require humans to operate them. AI operates within parameters set by humans, and will likely continue to do so for the foreseeable future. But within those parameters, it can often act independently—choosing, ranking, deciding. That alone makes it behave differently from land, machines, or even software in the traditional sense.

A factor of production isn’t just a philosophical label. It exists to help us explain and measure the economy. If adding a factor doesn’t improve growth accounting, policy design, or business strategy, economists won’t bother. This is why some researchers talk about “digital labour” or “machine intelligence”: not because they want new categories, but because too much productivity is currently being buried in the Solow residual—the box labeled “we don’t quite know where this came from.”

AI clearly enhances human productivity. That part isn’t controversial. In that sense, today’s AI can reasonably be described as technology—a powerful one, but still technology. It processes information created by humans and executes objectives defined by humans. Like other technologies, it raises output.

But AI also does something no previous technology has done at this scale. It automates parts of cognition itself. Even if it is only rearranging human-made information, no human can do so at its speed, breadth, or consistency. This is where the analogy with ordinary technology starts to strain.

If AI were simply capital, it would behave like other capital goods. It doesn’t. If it were just labour-saving technology, it would enhance labour without resembling it. It increasingly does resemble labour—except non-human, infinitely replicable, and made rather than born.

This is why I’m inclined to think AI may become a factor of production—not because it is “intelligent” in a human sense, but because it fits awkwardly into every existing category. I’m wondering if, when something doesn’t fit any of the existing buckets cleanly, maybe it deserves its own bucket. For now, AI probably still sits closest to technology: a tool that dramatically enhances output. But it is an unusual tool—one that changes the production function itself by substituting for certain kinds of cognition while amplifying others.

My next thought was what would happen if we did recognise AI as a separate factor. No country’s GDP would suddenly change; what would change is how we explain and decompose that GDP.

GDP today is built from three equivalent views:⁵ ⁶

Production approach: sum of value added = output − intermediate inputs
Expenditure approach: C + I + G + (X − M)
Income approach: sum of factor incomes (wages, profits, interest, rent) plus taxes less subsidies

All three are about the value of final goods and services produced in a period, not about how many “factors” are in the textbook. So just declaring “AI is now a factor” would not suddenly make India’s or any country’s GDP number jump.

In growth economics, output of an economy is often represented as a function of two primary, measurable inputs:⁷ ⁸

Labour
Capital

A standard production function can be written as:

Y = F(K, L, A)

where Y is income, K is capital, L is labour, and A is a catch‑all “technology” term—the Solow residual. If AI or “digital labour” became a recognised factor, you’d move to something like:

Y = F(K, L, A, D)

where D is an explicit AI/digital labour input, and A remains the residual technology that is not AI.

That doesn’t change the level of Y we measure as GDP, but it does change the story of where Y came from: part of what is now “mystery productivity” (TFP/Solow residual) would be reassigned to a measured AI input. In other words, the pie stays the same size, but we start saying, more precisely, which ingredient did how much of the baking.

Sources

Tag: AI

Can AI be a new economic factor of production?

Share this: