Science for Health keynote speaker Mieke De Ketelaere: The importance of ethical AI

MEDVIA and BioWin are pleased to welcome Mieke De Ketelaere as a keynote speaker at Science for Health 2024. Her talk, “Artificial Intelligence: Hindsight, Insight, Foresight”, will reflect on the original dream around AI, look at where we are 70 years later, and discuss what we need to understand to get it right in the future.

An engineer and artificial intelligence expert, Mieke started her career seeking to incorporate AI in healthcare and then moved swiftly into a consulting role. Mieke has more than 30 years of experience in business strategy and has held several management positions. She also has the distinction of being awarded Computable’s IT Person of the Year for 2024. Mieke currently works as a consultant and professor, teaching Ethical and Trustworthy AI at the Vlerick Business School in Brussels.

Q: What is missing in conversations about AI?

Mieke De Ketelaere: I’m an absolute believer in the value of AI, but everything I do involves a balanced approach in embracing the technology. It’s here to stay, but we don’t seem to be able to get a balanced view in the media. AI is currently in the hands of a couple of players who shape our vision in a certain direction.

A lot of companies – in Europe, the US and globally – are starting to invest in AI. But only 25% of companies have success with their AI products, as a quality increase or an efficiency gain in internal and external processes. The others see increased cost and decreased quality after putting their AI project into operations. I want to demystify AI in order to make sure that everybody understands how to create value with AI, while understanding the limitations and risks.

So what do you have to do as a company to not fear this technology? You start with a challenge or objective in mind. From there you analyze which data, insights and actions you need to setup to meet your goal. You do this in an iterative approach from day one, and measure after every circle in order to make sure you are on the right track with your AI system. And if you do this in a very understandable language at each of the steps, people feel critically equipped to understand where the value is and where the potential misunderstandings are. This is what I do.

“It’s important to think about where your technology is going to be used, and by whom.
And it needs to happen before you start working on it, not when the system has been activated.”

What is ethical AI and why is it important?

Ethical AI is understanding that these systems are learning from data. And in this data there are human mistakes linked to diversity and inclusion themes. The system by itself is not able to filter out these human mistakes.

We should understand that if you’re going to create systems based on statistics, that these systems will learn our mistakes and normalize them. We’ve seen this in all types of places, like the Apple Watch, which doesn’t work on dark skin. Human rights have to be a priority. So being critical of the system, being aware of potential biases. You have to think about the system as being human. That’s an example of the ethical part, but there are many more.

How do you implement ethics into AI development?

Well, the core of my work is to do it in a multidisciplinary way. You can’t expect engineers to have that knowledge – we weren’t educated for this. It’s necessary to create a multidisciplinary team that can provide different knowledge and perspectives so that we can reduce bias in certain decisions.

As an example: with new Mercedes Benz vehicles, you have to start the AI by saying, “Hi, Mercedes”, and then you can ask it to bring you to A, B, C or switch on something in my car. As a woman, a lot of times, I have to lower my voice to get the AI system to recognize me. This was likely unintentional by the engineers, but they are only human. It’s important to think about where your technology is going to be used, and by whom. And it needs to happen before you start working on it, not when the system has been activated.

“In data there are human mistakes linked to diversity and inclusion themes.
The [AI] system by itself is not able to filter out these mistakes.”

What are the benefits and concerns of using AI in healthcare?

The reason I studied AI was because they promised me at university that this is a technology that can reduce uncertainties in life. Health has a lot of uncertainties. Who is going to develop an illness? Who is going to react to this medication?

This is why I started my early career in healthcare. The research of my PhD was to create an AI system to predict when it was and wasn’t safe to induce labor in a pregnant woman with medication. Today, there’s a lot more health data which allows previously unseen analyses of patient trajectories or even molecules with AI technologies. Our brains can’t simply analyze this amount of data in a time-efficient way. AI can help us to lessen the uncertainties, to say: “Well, here might be potential cancer. I can see this from the past.” And so it’s technology that then helps us in the world of uncertainty.

Don’t miss Mieke’s keynote at Science for Health – register here!