Capabilities, Common Failures, and (Mis)Alignment with Ollie Jaffe (Open AI)
Penn Linguistics presents the second session of the three-part workshop series The New Mind. This session explores the capabilities and limitations of LLMs introducing insights into how these systems process information, make predictions, and sometimes fail. Participants will examine the relationship between reasoning and pattern recognition, learn to recognise what problems could be easy and difficult for an artificial mind, and discuss the risks of unexplainable AI in both technical and societal contexts.
Participants interested in computationally accurate and at the same time socially impactful research will have an opportunity to engage with capabilities- and alignment-related concepts that will empower them to integrate computational perspectives into their own academic fields in a meaningful way. All researchers—from aspiring scientists to senior scholars—as well as those simply curious about AI are welcome to join.
Vegan, kosher, and allergen-friendly food will be served at the venue.
The talk will commence at 3.30 PM but participants are very welcome to come in ahead of time and enjoy a cup of tea with snacks.
Learn more about this event series on our website.