Thijs Jochems: The AI paradox: a ‘smart’ economic stagnation?

October 17, 2025

^{This column was originally written in Dutch. This is an English translation.}

Wherever you look, you see news about artificial intelligence (AI). If politicians, economists and businesses are to be believed, a new digital dawn is upon us. Billions are flowing into start-ups that “do something with AI”, in the belief that we are on the eve of a new golden age. AI will deliver transformative efficiency gains. A nice promise, but will it be fulfilled?

By Thijs Jochems, Consultant and Private Investor

This expectation is not reflected in the productivity figures of Western economies. Productivity growth has been slowing since the 1980s. That growth – creating more value with the same input of people and resources – is the only true driver of sustainable prosperity. In 1987, Robert Solow formulated the “productivity paradox” as follows: “We see computers everywhere but in the productivity numbers.” Are we facing a similar situation with AI?

Many factors, including the metrics, play a role. It is well known that GDP is not an adequate measure of productivity growth. We will leave that aside for now.

The core of the paradox may not lie in the technology itself, but in its application. As with all innovative technologies, the return is not generated by the technology, but by its productive application. Here we see a crucial distinction.

The most important factor may be that AI is mainly used for activities that do not increase productivity. A large part of AI investment focuses on what we might call “displacement success”: influencing consumer behavior to gain market share. Think of armies of brilliant engineers perfecting algorithms—not to optimize logistics chains, but to keep our attention on a social media feed for a few seconds longer. This generates profits for one company but does not lead to an increase in social productivity. It is a zero-sum game within a stagnating economy.

Of course, other factors play a role, such as an inevitable implementation delay—comparable to the decades it took to fully integrate electricity—and chronic measurement problems in our service economy. But the most pressing question is whether we are focusing our capital and talent on the right problems.

The focus on non-productive applications of AI is not an abstract academic debate, but a direct threat to our prosperity and democracy. Western societies are facing an aging population. A smaller workforce means productivity must increase if we want to avoid a long-term struggle over the distribution of shrinking prosperity. Excessive inequality is fertile ground for polarization and populism. That is not good for the future of democracy.

This also has practical implications for pension funds and other institutions with obligations that will continue for decades. They depend on assumed returns based on expected growth, and their investments are built on this assumption. No growth means no returns. Adequate pension payments then become a social problem.

The conclusion is: be selective. Politicians and investors must carefully consider which AI activities and investments they want to promote or finance. It is their job to direct capital and policy toward applications that deliver real social productivity growth. Distinguishing between investments that enrich society and investments that merely maintain the status quo is crucial for our future. If we fail to do so, we are not investing in a richer future but merely financing a smarter version of economic stagnation.

More columns by Thijs Jochems