AI Can Now Adapt to New Tasks. Can It Also Improve Itself?
Large language models now adapt to new tasks, as the ARC-AGI-3 benchmark shows. In a new study, we asked whether they can also improve their own harness for this benchmark. They build harnesses well, but their improvements rarely transfer to games they have not seen. We discuss why and what that means for AI research.
Beyond Boundaries: A Human-like Approach for Question Answering over Structured and Unstructured Information Sources
Title Together with my co-authors, we are excited to share our work on an easy-to-apply method to improve LLM reasoning and how we applied it for question answering across heterogeneous sources. 🚀 [...]
The Next 66 Years of Artificial Intelligence
[Disclosure: The post below reflects a personal opinion and is not linked in any way to my employer. Sources for images and quotes are at the end of the article.] In 1956, a workshop took place at [...]
Neural Language Models – A Path to Artificial General Intelligence or a Blind Alley?
[Disclaimer: The text below represents a personal opinion based on the scientific evidence we have so far.] With the rapid progress AI research has made over the past years, the decades-old questio [...]
Website Update
Website Update Over the past weeks, my website underwent significant design updates and some restructuring. Content-wise, I added further information about the research work we are / I am doing. I w [...]
