Skip to main navigation Skip to main content Skip to page footer

BLISS AI Speaker Series: The Importance of Model Specs and Constitutions for AI Safety – Robert McCarthy

Research talk by UCL PhD researcher Robert McCarthy exploring the importance of model specifications and constitutions for AI safety, followed by networking with Berlin's machine learning community.

How can model specifications and constitutions contribute to the safety of artificial intelligence systems? As part of the BLISS AI Speaker Series, Robert McCarthy, PhD student at University College London (UCL), will present his talk “The Importance of Model Specs and Constitutions for AI Safety” at Technische Universität Berlin.

The approximately 45-minute presentation addresses a topic in AI safety and the development of advanced AI systems, focusing on the role of model specifications and constitutions. These approaches concern how intended model behavior, guiding principles, and safety expectations can be defined, raising important questions for the development of increasingly capable AI models.

Organized by BLISS Berlin e. V., the event is part of its winter 2026/2027 speaker series, featuring researchers from institutions including Google DeepMind, Anthropic, the Max Planck Institute for Intelligent Systems, Imperial College London, Microsoft, and UCL. Following the presentation, participants can exchange ideas, discuss research questions, and connect with fellow AI enthusiasts over complimentary drinks and pizza.

  • Date: January 12, 2027
  • Time: 7:00–9:00 PM
  • Location: Technische Universität Berlin, Institute of Chemistry, Straße des 17. Juni 115, 10623 Berlin
  • Organizer: BLISS Berlin e. V.
  • Speaker: Robert McCarthy, PhD Student at University College London (UCL)
  • Talk: The Importance of Model Specs and Constitutions for AI Safety
  • Registration: RSVP via Luma required to guarantee entry
  • Audience: Students, PhD candidates, academic researchers, industry professionals, and anyone interested in advanced AI research
  • Focus: AI safety, model specifications, AI constitutions, machine learning, and advanced AI systems
  • Program: Approximately 45-minute research presentation, followed by discussions and networking
  • Networking: Complimentary drinks and pizza
  • Important: Doors close at 7:15 PM; early arrival is required
  • Note: Participants agree to being photographed during the event

Newsletter

News, events and success stories from the Berlin AI ecosystem - once a month and directly to your inbox. Subscribe now!