idw - Informationsdienst
Wissenschaft
The psychologist and AI researcher Prof. Dr. Markus Langer explains in the following interview how AI manipulates us and threatens our democracy. At the same time, he specifies the opportunities a regulated use of AI could open up for our political system.
Prof. Langer, how will the use of artificial intelligence change our democracy?
Democratic processes can benefit from AI, because people are better informed, are better able to verify facts, and can use AI to put themselves into the shoes of people with opposing viewpoints and understand them better. This has been demonstrated in studies. But of course, there are also negative effects that we are seeing more frequently at the moment: AI is being used to spread fake news and manipulate opinions – at a speed no human can keep up with. AI is technology. I can design it in a way that either promotes or undermines democracy.
Could AI manipulate the shaping of public opinion in such a subtle way that we can no longer even make out the intention behind it?
That is de facto already the case. When we talk about AI, we often mean language models. They are extremely good at convincing us. They are not optimized to discriminate between truth and untruth but to sound plausible and to get us to keep speaking with them. It’s no problem to design an AI that explains why heat pumps aren’t a good idea, to take a harmless example. Such a language model will take up everything users say and recognize: I can convince this person better at an emotional level and that person better with arguments. Perhaps the AI will even make up fictitious studies in the background and then cite them itself. In this way, it attacks people’s hearts and minds.
And in the end, I feel understood and am convinced that heat pumps are utter nonsense ...
Exactly. The more I speak with such models, the better they figure out how to tell me what I want to hear. At the same time, many people regard AI models as virtually neutral machines. That makes them even more dangerous. Even on social media, there are little bubbles where attention is directed only towards content that aligns with one’s own opinion. But with AI, there’s a risk of creating a bubble that’s so small that it actually consists of only me and my AI.
And on top of that, there’s the risk of bias, depending on what data was used to train the AI model. Does AI reproduce social prejudices?
The data used is one thing: If I only use data from the USA and Europe, I can’t expect the AI model to reproduce the perspectives of sub-Saharan Africa. But the other thing is active human decisions in building an AI model, such as whether it is allowed to tell racist jokes or if limits are set on it.
Apart from all the risks that have to be taken into account: How could AI help us to make more rational, well-founded decisions?
It is possible to build an AI model in such a way that it actually helps us to make decisions. If I say, ‘I would really like to know what everyone suddenly sees in heat pumps: Help me gain a broader understanding of them, criticize my own opinions on them, and provide me with facts on the topic’ – perhaps then the AI could help me weigh arguments against each other. In the end, I can still decide against getting a heat pump, but not on the basis of fake news or resentment. I think that many people don’t just want to hear what they believe already but actually use AI as a learning partner to be better informed. But that depends on how you use it and requires an AI model that makes that possible.
Politicians themselves could also use AI …
We’ll see a lot more of that in the future. If it goes badly, we could, for example, see more and more laws drafted or edited by AI without thorough review. However, AI can indeed do many things better: It can be wise to have texts or data checked by AI. Or an agency’s chatbot could help people fill in a form step by step and in any language without requiring them to read 37 pages of dense text. There are many possibilities for using AI in a sensible way. But it requires a kind of magic triangle: well-prepared people, well-built and regulated AI models, and favourable conditions for deploying them – in other words, enough time to make good decisions.
Can researchers make a contribution here?
Yes, if we work together across the disciplines. As a psychologist, I’m well equipped to explain how AI affects people and why. On this basis, I can derive design recommendations. To bring together the three dimensions of technology, people and context, I collaborate with researchers from computer science, philosophy and law – for example, within the Collaborative Research Centre “Center for Perspicuous Computing”, where we investigate effective human oversight and control of AI systems. Another example is the TITAN project, in which we explore the conditions necessary for the successful use of AI in the legal system.
#Demokratie-Dienstag
Stress test for democracy: What matters now
Six researchers – six scholarly perspectives on what our democracy needs now: Each week until winter break, the University of Freiburg will post an interview on a current challenge facing democracy.
In conversation with experts from psychology, sociology, political science, and history, the series will address the following questions, among others: Why do people support positions that are hostile to democracy? How do economic factors influence the stability of democracy? What role does political education play? How does digital misinformation affect democratic processes? What impact does gender-based political polarization have on democracy?
The researchers will report on how their research can help strengthen democracy and share ideas from their research on what political measures could contribute to stabilizing the democratic system.
In addition, the University of Freiburg will post interesting insights into the interviews on social media each Tuesday under the hashtag #DemokratieDienstag.
More about Prof. Dr. Markus Langer
Prof. Dr. Markus Langer is professor for work and organizational psychology at the University of Freiburg’s Institute of Psychology. There he heads the Psy:Tech Lab, which investigates how AI can be designed in a responsible, transparent, and human-centred way, particularly at the workplace. His book Der KI bist du egal: Warum wir uns Maschinen anvertrauen – und was das mit uns macht (AI doesn’t care about you: Why we trust machines – and what that does to us) will be published next February.
More information on the Psy:Tech Lab: https://uni-freiburg.de/aopsy/
Contact
University and Science Communications
University of Freiburg
Tel.: +49 761 203 4302
E-Mail: kommunikation@zv.uni-freiburg.de
https://uni-freiburg.de/en/ai-researcher-ai-can-either-promote-or-undermine-demo...
Prof. Dr. Markus Langer
Copyright: Silvia Wolf / University of Freiburg
Merkmale dieser Pressemitteilung:
Journalisten
Politik, Psychologie
überregional
Buntes aus der Wissenschaft, Forschungs- / Wissenstransfer
Englisch

Sie können Suchbegriffe mit und, oder und / oder nicht verknüpfen, z. B. Philo nicht logie.
Verknüpfungen können Sie mit Klammern voneinander trennen, z. B. (Philo nicht logie) oder (Psycho und logie).
Zusammenhängende Worte werden als Wortgruppe gesucht, wenn Sie sie in Anführungsstriche setzen, z. B. „Bundesrepublik Deutschland“.
Die Erweiterte Suche können Sie auch nutzen, ohne Suchbegriffe einzugeben. Sie orientiert sich dann an den Kriterien, die Sie ausgewählt haben (z. B. nach dem Land oder dem Sachgebiet).
Haben Sie in einer Kategorie kein Kriterium ausgewählt, wird die gesamte Kategorie durchsucht (z.B. alle Sachgebiete oder alle Länder).