Interview with Aura Martínez: Can AI feel like we do?

Ein Interview von Antonia Hirschmann

Martínez is pursuing her Ph.D.  at the Complutense University of Madrid, where she analyzes how artificial intelligence (AI) interprets human emotions in film reviews. For the winter semester, she is joining h_da, where she is not only conducting research but also teaching in the MTS master's program. In this interview, she talks about her research, her path into science, and what brought her to Germany.

Why did you want to come here for the winter semester?

I come from Tenerife, and when you live in the Canary Islands, you only have one season — summer. It’s sunny and humid every day, so I really enjoy the windy and cold days here. I just love winter. The only thing I don’t like is that it gets dark very early — around six it already feels much later — but that’s really the only downside for me.

Why did you choose Germany, and specifically Darmstadt, for your research stay?

I was lucky! My boss in Spain put me in touch with Professor Lars Rademacher, who kindly offered me this opportunity. I really like Germany. I’m learning the language, and the people here have been very nice to me. Germany also has a strong focus on technology and excellent university programs. Darmstadt, in particular, offers interesting programs in cinema, AI, and computer science that perfectly match my interests. I hope to work here in the future.

How did you get into science?

I studied psychology, and I also have a master's degree in criminology. Initially, I wanted to be a criminal profiler. Then, I started studying journalism. But I also love science, and I’m a big fan of cinema and sci-fi. In my Ph.D. program, I’m bringing it all together. Science fiction movies allow us to dream about things like AI and machines. I also try to understand society, which is how I connect psychology and journalism in my work.

What is your research about?

I want to know if AI can feel like we do. My project analyzes AI’s ability to process human emotions through film reviews.

How does your research work?

I work in three steps. First, I analyze a film review myself and then give the same review to ChatGPT to analyze. Then I compare both analyses, the human and the AI, to see the differences. In the second part, I give the same review to ChatGPT again, months later, to see if it’s consistent in its responses or if it changes them. Finally in the third part, I study the types of words it uses to describe emotions.

Why did you choose film reviews as your data set?

Because the answers are not obvious. Reviews contain a wide range of emotions, such as happiness, sadness, loneliness, and empathy, all within a short text. Reviews can also be complex, using irony or sarcasm. If you used a romantic book, for example, you would only find emotions such as love or hate. In reviews, however, you get many emotions mixed together. 

Did some of your results surprise you?

Yes, mainly the lack of consistency. ChatGPT changes its answers after a few months. It also often uses nouns and verbs to describe emotions, which surprised me. For instance, it considers words like "authority" and "crisis" to be emotions, but they're not. This demonstrates how AI struggles to interpret the complexity of human emotion. I still have to finish the project, but these results already show that AI doesn't truly "understand" feelings the way we do.

What is important about your research? Who can use the results?

I want to show whether machines can be like us and understand emotions the way we do. Many researchers are trying to create robots that resemble humans, with brains, bodies, even feelings.  I hope my project can serve as a small step in that direction. If machines can understand our emotions, they can help us better, even by diagnosing illnesses or supporting mental health. AI is very important in the medical field. That’s why I believe this research matters.

How do you look at AI — as a threat or a chance?

I’m a big fan of AI and I see it as a friend. A lot of people think AI will destroy us, but I believe that, in good hands, AI can be our salvation. It can help humanity. We just need to use it responsibly. Maybe not in five or ten years, but in a hundred years, I think we’ll have robots like in the movie I, Robot or Wall-E. We'll have intelligent houses and emotional machines.