Ram Shankar Siva Kumar | A Few Useful Lessons about AI Red Teaming
AI red teaming is exploding in popularity: At DEF CON this year, more than 2,500 hackers descended to red-team AI systems. Every organization investing in AI—from Microsoft to Google to Meta to NVIDIA—has AI red teams to actively secure their AI systems
But what does it even mean to red-team AI systems? Grounded in case studies from Microsoft, Siva Kumar contextualizes how red-teaming AI systems differs from red-teaming traditional software systems, discusses how it intersects with previous lines of inquiry such as adversarial examples, and distills eight lessons from a practitioner’s perspective.
Watch Event Recording
Related Events
Ashesh Rambachan | From Next-Token Prediction to Automatic Induction of Automata
Sequence data is ubiquitous in economics — job histories in labor economics, diagnosis and treatment sequences in health economics, strategic interactions in game theory. Generative sequence models can learn to predict these sequences well, but their complexity makes it hard to extract interpretable economic insights from their predictions.
.png&w=1920&q=100)
Sequence data is ubiquitous in economics — job histories in labor economics, diagnosis and treatment sequences in health economics, strategic interactions in game theory. Generative sequence models can learn to predict these sequences well, but their complexity makes it hard to extract interpretable economic insights from their predictions.
Caroline Meinhardt, Thomas Mullaney, Juan N. Pava, and Diyi Yang | How Can AI Support Language Digitization and Digital Inclusion?
What does digital inclusion look like in the age of AI? Over 6,000 of the world’s 7,000-plus living languages remain digitally disadvantaged.
What does digital inclusion look like in the age of AI? Over 6,000 of the world’s 7,000-plus living languages remain digitally disadvantaged.
Matt Beane | Precision Proactivity: Measuring Cognitive Load in Real-World AI-Assisted Work
Systems like ChatGPT and Claude assist billions through proactive dialogue—offering unsolicited, task-relevant information. Drawing on Cognitive Load Theory, we study how cognitive load shapes performance in AI assisted knowledge work.
.png&w=1920&q=100)
Systems like ChatGPT and Claude assist billions through proactive dialogue—offering unsolicited, task-relevant information. Drawing on Cognitive Load Theory, we study how cognitive load shapes performance in AI assisted knowledge work.