On the scientific agenda
The first paper that the team published, has now been cited hundreds of times by scientists around the world. Van Harmelen comments: “Thanks in part to this publication, the topic has really gained momentum. More than ever, science is focusing on collaboration between humans and AI, and the technology that supports it, rather than simply asking how robots can replace people. We call this Hybrid Intelligence.”
In addition to conducting their own research, the team was also set up to foster international collaboration. To that end, they organized the at VU Amsterdam. Van Harmelen explains: “This has really taken off in Europe. After the first edition in Amsterdam, the conference has since been held in Munich and Malmö, with the next one scheduled for Pisa.” These efforts have led to an increasing number of partnerships and have established VU Amsterdam as a pioneer in the field of human-AI collaboration.
Engaging with societal projects
To support this research, Van Harmelen received a Gravitation Grant from the Dutch Research Council (NWO). The grant spans a ten-year period, and the team is now at the halfway point.
The team is working on a range of societal projects. Van Harmelen explains: “One of our applications provides lifestyle advice for people with diabetes and those around them. Most current apps that support diabetes patients focus solely on changing the patient’s behaviour, like taking more exercise, choosing what (not) to eat and at what times. But it’s better if this is done in consultation with healthcare providers (GPs, nurses, specialists and pharmacists) albeit without increasing their workload. Hybrid intelligence (HI) systems can help facilitate better and more efficient ‘dialogue’ between all these parties, including the patient. For instance, an HI system could help explain healthcare professionals’ advice more clearly to the patient.
“The team is also collaborating with microsurgeons at Erasmus University Medical Center in Rotterdam. These surgeons work with extreme precision and use a camera for assistance. They want to be able to control the camera using AI, so that it understands medical language.”
Further developments
The University of Groningen is experimenting with video games, whereby the team can set how much the computer knows about its opponent or teammate. The results show that the more information the computer has about its opponent, the better it can play the game. Van Harmelen says: “That turns out to be an important element in enabling computers to work more effectively with people. As researchers, we're using that knowledge to further develop hybrid collaboration with AI.”
A development that’s moving faster than expected comes from Google DeepMind. Google has created an AI program that can collaborate with scientists. is working on a similar application, says Van Harmelen. “Scientific research follows a predictable cycle: you conduct a literature review, formulate a research question or hypothesis, design and execute an experiment, analyse and interpret the results, and finally report them in a publication. AI can assist researchers at each of these stages. In some steps – like reviewing literature or analysing data – it already performs quite well. AI has also made great strides in recent years on supporting the writing of publications. However, AI still can’t formulate new hypotheses or design new experiments, but perhaps that will be possible in the future. At Hybrid Intelligence, we’re actively working to improve these AI tools for researchers.”
What the future looks like
Van Harmelen is cautious about predicting the future. “I’ve been working in this field for about 40 years, and I’ve never seen developments move so fast. It’s as if all the innovations are reinforcing each other, and we’ve found ourselves in a kind of whirlwind of progress. So I honestly don’t know what the future look like for human-AI collaboration. But I do hope that norms, values, responsibility and cooperation remain central to future developments. In America, the ethics teams have been disbanded in recent months under Trump’s leadership. And what China is doing in the field of AI is not entirely clear to the outside world. This gives Europe a great opportunity to take the lead.
“To make organisations future-proof and ensure that their digital transformation does not alienate people, but rather connects them, it’s important to educate the population. People must understand both the limitations and possibilities of AI in order to benefit from it fully. That’s why the Dutch universities have launched a national that’s free for everyone to follow.”
