Filip Sondej studied cognitive science and computer science. Over the past few years, he has been working on multi-agent AI safety. Currently, he is researching methods for removing dangerous behaviors from large language models (LLMs).
Adress: ul. Ingardena 3, 30-060 Kraków
• AI safety
• LLM unlearning
• Chain-of-Thought faithfulness
• AI conflict
• Digital sentience
• LLM unlearning
• Chain-of-Thought faithfulness
• AI conflict
• Digital sentience
NCN Sonata Bis 10 2020/38/E/HS6/00490 - The multidimensionality of indecision and its impact on cognitive control: A neurophysiological approach