Mathijs Bouman: is AI a lifeline for our aging population?

Bouman begins his story with perhaps the biggest challenge facing our country: the tight labor market. It is already under extreme pressure and will only worsen in the coming decades. Unemployment is historically low, while vacancies are abundant. “We actually need more people, but they are not there. Labor participation is already extremely high and bringing in immigrants is politically sensitive.” 

And then the biggest challenge is yet to come: aging. In 1950 there were still seven workers for every retiree, now there are only three. And that number continues to decline. This is not a temporary problem, but a structural shift that will continue for the rest of the century. His conclusion is unavoidable. “If we want to keep growing as an economy, our labor productivity must increase: doing more with fewer people.” 

According to him, we must let go of the fear that AI and automation will destroy our jobs. It is actually necessary. “I still hear local politicians saying they want to attract companies to their region to create jobs. But in a country with structural shortages, that is the wrong reflex. By organizing work more intelligently and automating it, people become available for sectors where the need is truly high, such as healthcare and education.” 

AI as our economical lifeline

To grow our economy, we need to increase productivity. More output per hour worked. And that is exactly where things have been stuck for years. In the 1970s productivity still grew by about four percent per year, now it hovers around zero. That is remarkable, Bouman argues, given the massive digitalization, automation and innovation of recent decades. “As the well known economist and Nobel Prize winner Robert Solow put it: we see computers everywhere except in productivity statistics.”

And that is what makes the rise of AI so interesting to him. Is this finally the technology that will break that trend? Will Agentic AI deliver the productivity boost economists have been waiting for for decades?

ChatGPT leads to ‘desk sludge’

In theory it looks promising. Experiments show impressive results. In a call center, AI led to 14% higher productivity. Consultants worked faster and delivered higher quality. “Those are numbers that make an economist enthusiastic. But as soon as you step out of experiments and look at the real world, the picture changes. Large studies among thousands of companies show something different. Despite significant investments in AI, there is hardly any measurable impact on productivity.”

In fact, in some cases AI even seems to have the opposite effect. Bouman refers to what in English is called workslop, which he translates into Dutch as ‘bureaubagger’ (desk sludge). “With tools like ChatGPT, even the most average employee can effortlessly generate texts. But they are often incorrect. Then you get what I call ‘desk sludge’.” And that has consequences. “If you want to send those texts to customers, they first have to be checked. What you see is that senior employees spend a lot of time correcting work that did not even exist before.” The result: productivity decreases rather than increases.

Breakthrough Agentic is yet to come

According to Bouman, however, this certainly does not mean we should write off AI. “History shows that major technological breakthroughs take time, sometimes decades. The steam engine, electricity, the internet; all of them took longer than expected before their real impact became visible.” GenAI and Agentic AI are still at the beginning of their development. “We are somewhere between the hype and real value creation. The technology is there, the promise is clear, but the real breakthrough is yet to come.”

Despite his critical tone, Bouman does not end his keynote on a pessimistic note. AI as a lifeline for our economy? Not yet. AI as a necessary development in a world with structural labor shortages? Without a doubt. “We simply need this technology. Not because it already works perfectly, but because there is simply no alternative.” 

* This blog is based on the keynote by economist Mathijs Bouman during IncentroCon Agentic '26.