"We are virtually creating a new ecosystem, which is based on radical openness."

Health is a precious commodity and yet access to life-saving technologies and procedures is, to a great degree, unevenly distributed around the world. But what if everyone had access to the same tools and resources needed to diagnose and treat cancer? The Hippo AI Foundation is taking a radical approach and wants to make medical knowledge gained through artificial intelligence freely available by means of an open knowledge licence. We spoke with the two founders Bart de Witte and Viktoria Prantauer about the beginnings of their groundbreaking idea, the hurdles of de-economising data, and the importance of collaborations.

First of all, congratulations on winning the Startup Award of the German AI Prize! Did you expect this and how did you celebrate the win?

Victoria: So I think I’m still celebrating and on my part I can say, for me there was only this option: If we already get this opportunity to be there, then we have to win that.

Bart: I was surprised because we were after all a non-profit, compared to two traditional for-profits. Accordingly, I was very pleased that they are also open to somewhat more sustainable approaches. It's a terrific recognition because we don't have such a simple vision after all. We are swimming against the tide and saying that we are demonetising data, while everyone else wants to monetise it. For us, this was definitely a milestone that helps us further and builds confidence.

Victoria: With the Hippo Foundation, we want to put society and the needs of society right at the forefront, and then to see – once again in this live voting – that this is exactly what society actually wants is wonderful. It was another indication that we are on the right track.

Let's jump back a few years: How did the idea of using open global datasets to train AI to improve cancer diagnoses come about?

Bart: This came out of my experience at IBM in machine learning. I've been thinking quite hard about what the long-term consequence is if we transfer the business model of data monetisation to medicine, and once again I've been thinking quite hard about the role of Europe. Are we even able to keep up in this race, since we don't have a scalable market? We don't have the conditions that China has or that the U.S. has, and, up to now, we have never managed to be a leader in this area of algorithms and data-based platforms. Google Search has always had a 92 percent market share in Europe for over 15 years, leading me to ask the question: What is the consequence when a platform has as much market share as Google Search? This is, after all, an algorithm-based business model. What kind of healthcare system are we in then, when all life-saving knowledge is monopolised?

Not a rosy future for global healthcare ...

Bart: Absolutely. The alarm bells went off for me, because a lot then also has to do with business models and morals. In America, the morality when it comes to health is completely different than ours in Europe, and I then thought that AI actually gave me a lot of hope that, all of a sudden, one could distribute expert knowledge to all corners of Africa. Yet, I see the opposite path being taken right now – towards blatant information asymmetries, where capital defines who can collect the most data and then generate IP on the algorithms.

How can this be counteracted?

Bart: I believe that, although politicians always mean well, they are not able to solve such problems. That is, I have had no confidence in politics, I have had no confidence in our market systems, because we don’t have the prerequisites. A purely market-based course, where the market defines the future – I don't believe in that either. I then was inspired by open source movements in software. I did a great deal of interviews and then tried to create something that is now the Hippo AI –  a vision, a narrative, where people can find each other and start working together on this basis, as open source is nothing else than a collaboration based on shared values and a framework.

How exactly does your licensing model work?

Bart: The idea is that we collect large reference datasets that are needed for AI modeling, and then provide them with a stamp, that is, an external licence notification. All licensees of such data are then obligated to always share it, and to put all derivatives under the same licence. In this way, we are virtually creating a new ecosystem, which is based on radical openness.

Certainly, such a task requires a great deal of expertise. Viktoria, you bring a personal story to Hippo AI, in addition to digital know-how.

Victoria: I come from the digital world, having built companies, startups and communities. For a long time, AI for me was just like, yes this is a tool, this makes us more efficient, helps companies become even faster and bigger – and that was my naive perspective on it. This perspective completely turned around once I was diagnosed with breast cancer in the summer of 2019. You have to imagine it like a shock wave running over you.

So I asked myself – for what do I want to use my knowledge and energy in the future? At the same time, one develops a great deal of compassion for all the others out there with the same pain. I then felt the need that now I just have to use this experience. This diagnosis and this treatment that I was able to get because I live in a country like Germany and simply have this access to all these options – I would like to make that possible for all other people.

How did you two find each other?

Victoria: Taking on that mission, I went on a search – and exactly three days after my diagnosis, I found Bart and was highly impressed with his approach – that we would use technology for the people and break the system of global injustices. A few months after my chemotherapy treatment, we met and together developed the concept of patient-driven projects.

In general, is there a willingness to make data available so that it truly benefits the public good?

Victoria: Communication with patients is extremely important, because this concerns our ability to, as people and as a society, design a different world. I believe, and this is also supported by corresponding research, that people are willing to make their data available if the output is commensurate. This means that if you ask those in the hospital, are you willing to make your data available for research, then most of us would tick yes. Yet, the problem is that, at that point, the question is not asked: What kind of research is this, anyway? Who benefits from the output of such research? If the patient knew this at the beginning, would they still consent? This discussion is missing.

Bart: I think that's something that's always so hidden. So then people always say: "Please make your data available for research" – but they don't then say that the research will lead to public knowledge. I think "Please donate your data for the common good" is a better approach, because you also have to make sure that the know-how is available to the public.

If a patient in a public hospital signs a document stating that he will participate in a research project and the data will then be sold exclusively to a pharmaceutical company, and that IP will be generated from it, then the patient thinks he is doing something good – but if the cancer therapy ends up costing 400,000 euros, then that has nothing to do with the common good.

Collaborations and partnerships are at the forefront of your project. How is your idea being received and who are your advocates?

Bart: I have noticed that the main supporters for our Hippo AI Foundation are the doctors and the patients. I've been fortunate enough to be able to speak as a keynote speaker at medical conferences, where I've been approached by professors and research doctors. Our message fits perfectly into the value system of a doctor: They need open access to knowledge, because they have actually always cultivated this philosophy. Without open access to data and algorithms, which in a sense maps future knowledge, doctors are disempowered. Of course, this is also a means for the doctors to continue to have control over the know-how, but also for our society.

Your mission is to de-economise data and AI in healthcare. To what extent do economic interests stand in the way of your goals here, and how do you counter your opponents?

Bart: What we're doing is actually something that a lot of people don't want to see, because they're betting that they’re going to monetise the journey or win. Yet, there is one thing that can’t be changed: This is the fundamental right to self-determination, and that means that we as human beings can decide to whom we donate or give our data. I think that, if we drive an open source strategy, then the economy will flourish a lot more, only we'll be able to do that among a lot more companies, instead of just single platforms that centralise everything.

Does Berlin offer a good basis for such a large-scale project?

Victoria: Personally, I came to Berlin because it's a gateway to the world, where a wonderful mix of people come together. This cross-fertilisation from all these different perspectives. That's what makes it so exciting and appealing, because you're not confined to your own bubble, where you’re trapped and unhappy – rather, Berlin offers you all these perspectives and allows you to see things from completely new angles.

Bart: Yet, I always find it difficult, because we have a global mission – and there is always this nationalism or this "we are now located here." Our goal is to build a global network that is highly decentralised and self-organised. But of course, the mindset of diversity, of social commitment, is already strong here in Berlin, so our goal fits very well here.

To be sure, open data projects stand or fall with cooperation. Recently you have been able to win over AstraZeneca, a major company, for your idea. How did that happen?

Bart: We were delighted when we were able to secure the company as a sponsor for the breast cancer project, and the willingness arose to invest in something that would then also be available to AstraZeneca's competitors. After all, that's what open source is. That's then available to everyone, but they understood that in such an exponential age as this one, the need for transparency and trust is becoming extremely important. One can see that many pharmaceutical companies now have no confidence at all among part of the population, and you can see that with all these discussions about vaccination. This means that they have an understanding of this, and are first testing this project with us to see whether these investments by AstraZeneca in open innovations will lead to faster innovations and more data – for less money and better outcomes. I hope that many others will follow.

Thank you for speaking to us.