Physicist
AI Aligner
Aligned AIs drive positive impact.
Philipp Alexander Kreer
Interdisciplinary Ph.D. researcher in theoretical physics at the Technical University of Munich and APART Research Fellow in AI alignment.
I commit my work to advance science and to improve our lives by aligning AI systems with human goals and values. I carry out my mission with:
Integrity: Taking responsibility, staying true to my values.
Openness: Embracing diverse disciplines, cultures, and criticism.
Fantasy: Pushing boundaries to discover innovative solutions.
AI Alignment Research
The strategic underperformance of AI systems during testing, known as sandbagging, poses significant security risks. For instance, an AI might conceal its cyberattack capabilities to evade governmental regulations. Malicious actors could exploit these capabilities after deployment. I developed a novel technique to detect sandbagging in modern AI systems that grounds in a mathematical understanding of AI systems.
Read Paper Inspect Code
What is the origin of mass?
I explore the nature of mass through two perspectives: particle collisions and gravitational waves. At CERN's Large Hadron Collider, we investigate how particles acquire mass through their interaction with the Higgs boson. Meanwhile, gravitational waves from black hole collisions provide insights into how mass behaves on cosmic scales. By calculating Feynman diagrams, I predict the outcomes of these experiments, connecting discoveries from the smallest particles to the largest structures in the universe.
Contact
Ready to start your next project with me? Questions on my paper or code? Or just a chat? I am happy to answer your LinkedIn message and E-Mail!
The two project pictures were generated with GPT4o. The jekyll template can be found here.