Adapting artificial intelligence to human needs
Pierre-Alexandre Murena is a new professor at TU Hamburg
13.09.2023
"My vision in dealing with artificial intelligence (AI) is to consider the human and the AI system as a team striving for a common goal. They should see themselves as partners in dealing with each other, giving each other hints, correcting each other and exchanging relevant information." This is what Pierre-Alexandre Murena, new junior professor at TU Hamburg, says about his work. It aims precisely at the interface between machine learning in its current form and its users. The mathematician and computer scientist deals with human-centred artificial intelligence. What everyone is talking about today still had an exotic status ten years ago. Murena reports: "When I mentioned my interest in AI back then, people asked me about Terminator and science fiction scenarios. Today, they tell me about their attempts to use AI like ChatGPT or Midjourney. This is a huge development, and I think it's important to guide people through it." Murena's goal is to make AI more human-centric, to make it easier for scientists and engineers who need machine learning models for their research to use AI tools. All of this is extremely complex in practice: because interactions tend to be brief and the AI is not given enough time to learn anything; and because cognitive models are complex and learning their parameters is often difficult to impossible.
"Training the AI is sometimes frustrating"
In one respect, humans differ from AI: humans interpret what others say, thus statements can become inaccurate or even wrong. "Adapting is an ability we have, but it is something AI lacks. A human-centred AI is supposed to close this gap one day. But the path to get there is sometimes frustrating because it seems so simple but is difficult to achieve!" says Murena. "Although I think the issue of human-centred AI will be a major challenge in the coming years, the focus of most universities and research centres is still on developing strong autonomous machine learning systems. The fact that TU Hamburg has established a professoral position on this topic was a good sign for me that the university anticipates this important change."
AI for the Arts
Prof. Murena studied Applied Mathematics and Computer Science at the Ecole Polytechnique and the Ecole Normale Supérieure in Cachan (France), and he obtained his PhD from Université Paris-Saclay. He then moved to Finland for four years as a postdoc at Aalto University and the University of Helsinki. There he worked on various machine learning projects with humans and led a research team at the Finnish Center for Artificial Intelligence (FCAI). "I have always been interested in science, especially mathematics. But I have another passion, which is art," says Murena. He tries to combine both and supports artists to work with AI. Outside of university, Murena can be found in the harbour area, where he photographs old industrial spoils. He also enjoys going to the cinema, playing the piano, cooking or preparing delicious flavors with his ice cream machine.
TUHH - Public Relations Office
Elke Schulze
E-Mail: elke.schulze@tuhh.de
