Evaluating and Controlling LLMs: Societal and Business Insights by Tutke and Root Signals
Tutke, The Finnish Center for Safe AI and Root Signals welcome you to an evening of discussion on the frontier of controlling and evaluating LLMs in Helsinki. There will be limited slots available.
Program:
18.00-18.10: Opening remarks and introducing speakers
18.10-18.50: Olli Järviniemi (Tutke->UK AI Security Institute) — Highlights from the Field of Evals
18.50-19.20: Snacks, refreshments, and open discussion
19.20-20.00: Root Signals — Practical LLM behavior steering: From Agent Evaluation to LLM Judge tuning
20.00-21.00: Free-form discussion and networking
Programming will be in English.
About Tutke
Tutke is an interdisciplinary non-profit organisation. We promote scientific understanding of secure development of artificial intelligence and support research and regulation leading to it.
About Olli Järviniemi
Olli Järviniemi is a researcher who specializes in AI control and deceptive alignment. He has a PhD in Mathematics from University of Turku, and he is shortly starting at the UK AI Security Institute.
About Root Signals
Root Signals helps companies speed up their adoption of GenAI by offering tools to measure, control, and evaluate LLMs. Root Signals lets you create, optimize and embed automated evaluators into your code, so you can continuously monitor the behavior of LLM automations in production.