USC Information Sciences Institute Courses

Why Are Human Laws So Difficult for AI to Follow?

Explore the challenges AI faces in interpreting human laws and regulations. Gain insights into open-texturedness and its impact on normative rule systems for agent-level AI development.

• YouTube
• 57 minutes
• On-Demand
• Free Video

OATH-Frames: Characterizing Online Attitudes Towards Homelessness with Language Model Assistants

Discover how large language models assist social work experts in analyzing millions of Twitter posts about homelessness through OATH-Frames, revealing crucial public attitude trends and policy insights.

• YouTube
• 1 hour 2 minutes
• On-Demand
• Free Video

Can LLMs Generate Novel Research Ideas? - A Study of AI Research Generation and Expert Evaluation

Explore groundbreaking research comparing LLMs with expert researchers in generating novel ideas, examining the potential and limitations of AI in accelerating scientific discovery through empirical evaluation.

• YouTube
• 57 minutes
• On-Demand
• Free Video

Why Larger Language Models Struggle with Subjective Reasoning

Explore how larger language models handle subjective tasks, examining their limitations in emotion and morality recognition despite advanced prompting techniques and increased model sizes.

• YouTube
• 57 minutes
• On-Demand
• Free Video

Do We Need Large Language Models for Time Series? - Understanding LLMs' Limitations in Time Series Analysis

Explore the limitations and potential of Large Language Models in time-series analysis, examining their effectiveness compared to traditional methods for applications in finance and healthcare.

• YouTube
• 47 minutes
• On-Demand
• Free Video

Weakly Supervised Learning for Adaptive Language Learning Model Agents

Explore cutting-edge techniques for training adaptable LLM agents through weakly supervised learning, focusing on frameworks like Agent Lumos and innovative approaches to enhance generalization capabilities.

• YouTube
• 1 hour 7 minutes
• On-Demand
• Free Video

Probing Covert Racism in Language Models Through the Lens of Dialect Prejudice

Explore how language models exhibit covert racism through dialect prejudice, focusing on stereotypes about African American English and their implications for AI development and social impact.

• YouTube
• 1 hour
• On-Demand
• Free Video

Mission: Impossible - Testing Language Models on Unnatural Grammar Rules

Explore groundbreaking research challenging assumptions about language models' ability to learn impossible languages through systematic experiments with synthetic language variations and GPT-2.

• YouTube
• 58 minutes
• On-Demand
• Free Video

AI Trustworthiness Index and Sources Monitoring

Explore the importance of AI trustworthiness and source monitoring, and learn about knowledge graphs, AI assistants, and potential impacts on critical thinking.

• YouTube
• 20 minutes
• On-Demand
• Free Video

Towards Socially Aware and Safe AI Agents

Explore groundbreaking research on developing socially intelligent and safe AI systems, focusing on LLM evaluation frameworks, information asymmetry challenges, and safety considerations in human-AI interactions.

• YouTube
• 1 hour 1 minute
• On-Demand
• Free Video

Duality of Bias in Large Language Models - Leveraging Community Perspectives and Uncovering Ideological Vulnerabilities

Explore the dual nature of bias in Large Language Models through research examining community perspective alignment and ideological vulnerabilities, with insights for responsible AI development.

• YouTube
• 1 hour 6 minutes
• On-Demand
• Free Video

Harmful Speech Detection by Language Models Exhibits Gender-Queer Dialect Bias

Explore how content moderation systems and language models exhibit biases when detecting harmful speech in gender-queer dialects, with focus on reclaimed LGBTQ+ terminology and fairness implications.

• YouTube
• 58 minutes
• On-Demand
• Free Video

Designing Priors for Bayesian Neural Nets to Enhance Probabilistic Predictive Models in Engineering

Explore how Bayesian neural networks can enhance engineering predictive models by integrating physics-based knowledge and providing robust uncertainty quantification for high-consequence decision-making applications.

• YouTube
• 1 hour
• On-Demand
• Free Video

We Over Rely on Technology but Lack the Resilience We Need to Depend on It

Explore cybersecurity challenges with David Balenson as he discusses autonomous vehicle safety, AI lessons from cybersecurity, and the paradox of our technological dependence versus resilience.

• YouTube
• 20 minutes
• On-Demand
• Free Video

Reimagining NLP Practices Towards Justice - From "Democratization" to Personal Names

Explore critical issues in NLP practices, examining the misuse of "democratization" and ethical problems with associating personal names with sociodemographic attributes, while offering recommendations for advancing justice in the field.

• YouTube
• 57 minutes
• On-Demand
• Free Video