Responsible AI and Data Management

The field of artificial intelligence and data management is moving towards a more responsible and ethical approach. Researchers are focusing on developing frameworks and techniques to ensure fairness, safety, and accountability in AI systems. This includes designing datasets that impart advanced capabilities to AI foundation models while minimizing the risk of AI misuse.

The importance of data repurposing is also being recognized, with a need to study and understand its implications for organizations and society. Furthermore, the trade-off between certainty and scope in symbolic and generative AI is being explored, with a conjecture that formalizes this trade-off and reframes engineering ambitions and philosophical expectations for AI.

Noteworthy papers in this area include: FAIRTOPIA, which introduces a fairness-by-design approach to disrupt unfair AI pipelines. The Alignment Trap: Complexity Barriers, which establishes fundamental computational complexity barriers to verifying AI safety as system capabilities scale.

Sources

Understanding and Improving Data Repurposing

FAIRTOPIA: Envisioning Multi-Agent Guardianship for Disrupting Unfair AI Pipelines

Situated Bayes -- Feminist and Pluriversal Perspectives on Bayesian Knowledge

Multiverse Privacy Theory for Contextual Risks in Complex User-AI Interactions

A Conjecture on a Fundamental Trade-Off between Certainty and Scope in Symbolic and Generative AI

Towards Responsible AI: Advances in Safety, Fairness, and Accountability of Autonomous Systems

Data-Centric Safety and Ethical Measures for Data and AI Governance

The Alignment Trap: Complexity Barriers

Built with on top of