Are the economic effects of AI oversubscribed?

Artificial intelligence
"We're exaggerating the dangers"

Christin Schäfer is one of the most important data pioneers in Germany. As a member of the Federal Government's data ethics commission and managing director of a data science company, she is convinced that the trend towards the data society cannot be stopped - but it can be shaped for the benefit of all.

Is humanity abolishing itself with the help of artificial intelligence?

I do not share this concern. I think that exactly the same will happen as in previous industrial revolutions, something is shifting. New tasks will arise, others will disappear. But man is by no means abolished. Above all, I see the development as an incredible opportunity. At work, we could focus on what you need brainpower for and delegate routine activities to the machines.

Can you understand that others are afraid of data freezing?

Yes, of course I can understand that. Still, I think we talk too much about the dangers and too little about the opportunities. And also exaggerate the dangers. It probably helps that I've learned to deal with data. That disenchants and demystifies things.

Your start-up acs plus promises: "data with care". How do you resolve the tension between what is technically feasible and what is legally permitted or ethically justifiable?

Much of what data-driven technology can do for us is completely devoid of any personally identifiable information. When I look out the window, it's nice and mild. But it's only the beginning of March! Climate change cannot be ignored.

But how do we manage to minimize energy consumption and lower pollutant emissions without our prosperity suffering significantly?

As a developed economy, we also need data-driven systems for precisely such complex questions. At acs plus we are currently working on these and similar technical issues. For example, I would never model something with the help of which supposedly certain people, ethnic groups or sexual orientations could be selected from photos or data sets.

Could it be that a large part of the diffuse rejection of the new data culture is due to the fact that it is simply being sold poorly?

I am happy to sign that. The engineer who has just ensured that an industrial plant emits 30 percent fewer pollutants thanks to a clever control system, rarely publishes blog posts. He's happy in silence. We need some positive anchor areas that also appeal to people's emotions. Maybe we should turn the tables and think: What would we not have if data technology didn't exist? How was it 15 or 20 years ago when you were on vacation without a smartphone, without a weather app, restaurant guide or sat-nav?

Who decides between “good” data services like these or “bad”? If politicians don't know what to do on such issues, they set up a working group. You are a member of the Federal Government's data ethics commission. What exactly are you doing there?

The 16-person committee examines the trinity of algorithms, AI and data - in order to clarify the crucial question: What can positive things be promoted with them and how negative things can be avoided? The demarcation should be based on ethical guidelines. When we look at the social credit system in China, we see the frightening counter-model: not everything that is technically possible and will become possible may also become reality. The package of tasks that politicians have given us to shoulder is enormous. We will present the final report in autumn.

In the free economy, data have long been ruling - in public administration, on the other hand, the analogue spirit of yesterday still prevails. Can algorithms help with better governance and administration?

I mustn't anticipate anything. Just so much: We are currently working intensively on this question in the Data Ethics Commission. It is one of the most exciting questions of our time.

What do you think of “coding” becoming a compulsory subject in schools?

Programming is not the most important thing in the data context. Good knowledge of mathematics and statistics is essential for the correct handling of data. The economy is developing in great strides towards machine learning processes. Many of these machines don't say yes or no. They say zero point seven percent or 90 percent. What do i do then? This aspect of digital competence cannot be achieved through programming; it is achieved through a solid mathematics education and, to a certain extent, your own data sovereignty: data sovereignty requires a playful, knowledgeable handling of data.

How do you deal with your own data in mobile or web activities?

I have a natural protection with social media because I just never really cared about it. In addition, I am already thinking about where to leave my data and where not. My private strategy, which is really not recommended, is simple: Everyone gets a little bit.

Christin Schäfer, born in 1974, has a common thread in her life so far: the love of data. Even in her childhood days in Kassel, she liked nothing more than painting the digits in the phone book with colors. After studying statistics with a minor in physics and specializing in computer science in Dortmund, she conducted research at the Fraunhofer Institute and worked in management roles at Deutsche Bank and an Austrian financial institution for around a decade. Today she is the managing director of her own company in Berlin: acs plus, a "boutique for data science".