Cognition, Agency & Generalisation in LLMs, by James Fodor

For our November event, we'll be joined by polymath and science communicator James Fodor, who is perhaps best known internationally for his "Science of Everything podcast". James currently specialises in computational neuroscience, researching the cognitive capabilities of LLMs at the University of Melbourne. During this presentation, he'll be covering:

• Why LLMs hallucinate
• Vulnerabilities and adversarial attacks
• An introduction to the concept of 'compositionality'
• Differences between human vs. LLM reasoning
• Cognitive architectures for LLM-based agents

Event details:

• Doors at 6:00pm
• Food + networking at 6:30pm
• Presentation ~7:00pm
• Q&A ~7:45pm

Places are limited, so be sure to RSVP!

To stay updated on future events, sign up for updates: https://eepurl.com/iuqHgY