Benjamin Kille on AI evaluation and innovation

Dec 17, 2024 | Interviews

As part of our TrustLLM project, we are happy to introduce Benjamin Kille, Associate Professor at NTNU’s Department of Computer Science. With a keen interest in AI trustworthiness, Benjamin is dedicated to advancing the field and helping Europe compete globally.

How do you think the TrustLLM project could impact the AI field?

The impact of any research project is often hard to estimate in advance. We see that there is a growing demand for answers to questions related to the trustworthiness, ability, and impact of generative AI. I hope that our efforts will help to answer these questions. Besides, I wish that the models that we will build together can help Europe to get more competitive with the US and China.

Which challenges do you find most interesting to solve through the project?

Personally, I am ranking the questions about evaluation the highest. Even after years of thinking about it, I cannot come up with an objective way to compare generative AI models. Selecting questions always seems to add a level of subjectivity.

What are you most looking forward to achieving in the coming months?

The first year has mainly served us to get set up. I am really looking forward to the first set of models being available and possibly aligned with the values that we want them to reflect. We have defined several exciting applications in which we want to test the models. I hope these models can make a difference and help us innovate beyond what current models allow.

What new insights have you gained from being part of this research project so far?

I find it extremely exciting to be part of a large consortium of researchers working on such a timely matter. On the one hand, it is exhilarating to see how many of our PhD students push the boundaries. On the other hand, it is reassuring to see that everyone is struggling with the same problems. If you work in your research group, you occasionally think that others have figured out things that you cannot seem to get through.

If you were to describe the project in one sentence, what would it be?

A very, very long sentence with many subclauses, as is necessary for Germans 😉