AI as “intermediate good”— with a twist

Kenneth Cukier
the self-driving company
2 min readAug 28, 2016

Artificial intelligence isn’t the end-product but a vital ingredient, even if we don’t yet know what those goods and services will be

IN THE late 1990s when cryptography was liberalized, I reckoned it would unleash a wave of crypto firms. A friendly guru in the field looked at me askance and explained that no one “bought encryption.” Rather, they bought point-of-sale credit-card readers, missile-guidance systems, email security…

Asphalt is boring, cars are cool

We are at a similar juncture with AI. There are firms aplenty. After all, techniques like deep learning as still embryonic, so there’s room for research-oriented companies. Think DeepMind (pre Google acquisition) or Geometric Intelligence: commercial players who aim to advance the science, not sell products.

It is a reminder that AI isn’t really a business in its own right, it supports other businesses. There is manufacturing. Healthcare. Banking. Law. Insurance. Education. Transport. AI will have profound effects there and everywhere. But its role is an intermediate good, not a finished good.

Yes, the same might be said of computers or software. But there is a consumer element to them: they’re directly used by people. AI is more like crypto: it enables things; it isn’t the thing itself. People want fast airline ticket searches, not algorithms. They want inexpensive, quick, accurate medical tests, not sigmoid functions.

A computer chip? Not yet. Just sand

AI is the asphalt, not the automobile. It’s the sand that goes into the microprocessor, that goes into the computer, that goes into the bank, that processes your transaction. As the legendary Harvard Business School professor Theodore Levitt recounted: people don’t want to buy quarter-inch drill bits; they want quarter-inch holes.

The interesting part, though, is that AI won’t just enable what we already do today but help it happen more efficiently. Instead, it will create the conditions for entirely new things that we can scarcely imagine.

This is what makes the current environment so fascinating. The AI community is developing new techniques to meet current needs, tailored around how the world works today. But the technology will give rise to new activities altogether, and require further development to fulfill needs that have yet to be created.

A new commercial sector is emerging to provide intermediate goods — when it doesn’t know what the end-products will be.

--

--