Watch the interview
This conversation of nearly ten minutes was recorded at the Kreativmesse fair. You can watch it here or open the original in the PAF-TV media library.
The interview is spoken in German with a Bavarian dialect. English subtitles are enabled by default. The video does not start automatically; you can change the subtitle language in the player.
Watch the original interview at PAF-TV
When an old photograph becomes a familiar face again
An old family photograph preserves more than an image. It reminds us of people close to us. In my own experiments with AI-assisted image reconstruction, I used historical family photographs, including small, cropped and damaged pictures of my aunt, my uncle and my father as a child.
I saw how remarkable the results can be, but also how easily the details of a familiar face can disappear. An image can look technically excellent and still fail to depict the person you know. Careful comparisons and switching AI models eventually brought one reconstruction closer to the original.
Taking AI’s limitations seriously
This experience also informs my work as an AI Agent Builder. A convincing result is not necessarily a correct one. Whether working with photographs, service agents or technical processes, what matters is that results can be traced and checked, with people able to intervene when needed.
My interest in AI is not limited to what is technically possible. I want to understand where it genuinely helps people and what responsibilities come with using it.
My thanks to pafnet.tv and Rolf Berger
I would like to thank the team at pafnet.tv for the opportunity to have this conversation. My particular thanks go to Rolf Berger, whom I have known for several decades. It was especially rewarding to talk with him about artificial intelligence at the Kreativmesse fair.
The team were volunteering their time at the fair, conducting interviews and presenting their own work. Such commitment deserves recognition. I hope this interview also introduces more people to what pafnet.tv does. Thank you for your time, the open conversation and your work at the event.
More about my work
Elsewhere on this website, I explain how I work and present selected AI projects. If you would like to discuss these topics, I would be pleased to hear from you.
Editorial companion text, not a verbatim interview transcript. Source: PAF-TV media library.