Healthcare organisations also need to invest in new skills, different working methods and room to experiment. Volberda, professor of Strategy & Innovation, and De Mast, professor of Business Analytics, co-authored the article with Henk Marquering of Amsterdam UMC and ARCHEM director Annamarike Seller-Boersma. They explain why limited productivity growth does not necessarily mean that AI’s potential is exhausted.

More than improving existing processes

The authors distinguish between incremental and radical innovation. Incremental innovation improves existing care processes step by step. Radical innovation makes entirely new ways of working possible. AI’s greatest contribution may therefore come from changing how healthcare is organised, rather than simply making current tasks faster.

Remote patient monitoring illustrates this distinction. Initially, it offered a way for patients to take measurements at home instead of visiting a hospital. Over time, it also made a different approach to care possible: continuous monitoring, with healthcare professionals intervening when medically necessary rather than relying on scheduled check-ups.

Discovering such possibilities requires learning by doing. Organisations, professionals and patients need to explore which new processes work and how to use them effectively.

Why productivity gains take time

New technology does not immediately translate into higher productivity. This is known as the productivity paradox. According to the authors, value emerges when technological advances are accompanied by investment in skills, redesigned processes, better data infrastructure and new forms of collaboration.

These changes initially take time and capacity. Staff must learn to use new systems, while organisations experiment with applications whose benefits remain uncertain. Measured productivity can consequently fall before it rises. The authors describe this pattern as the productivity J-curve.

Room to experiment safely

Cost-effectiveness remains the goal, the authors emphasise. However, judging every early AI application solely on proven cost-effectiveness risks stopping promising innovations before they can demonstrate their value.

They argue that the Dutch healthcare system makes insufficient distinction between gradual improvements and radical innovation. Healthcare leaders should therefore create safe opportunities to experiment, learn and adapt alongside their investments in technology. The largest productivity gains may come from discovering fundamentally different ways to deliver care.