Evžen Wybitul

Evžen Wybitul

MsC Student

ETH Zürich

Evžen (/ˈɛvʒɛn/) is a Data Science student focusing on AI Safety topics. He is Czech and did his undergrad in bioinformatics in Prague. Recently, he has been a part of the MATS program in Berkeley, where he worked on scalable oversight solutions for RL. Next, he would like to do research in mechanistic interpretability, at least in part hoping that all the cell biology and life-science methodology courses he had to take would become useful. Most of his reading time is thus taken by interpretability papers; he has also been recommended to set some time aside to read up on the sunk-cost fallacy.

(fun fact: the caron above the ‘z’ in my name is called a haček (/ˈɦaːtʃɛk/) in linguistics, a term directly taken from the Czech language.)

Interests

  • AI safety
  • mechanistic interpretability
  • scalable oversight