Responsible AI for General-Purpose Systems: Overview, Challenges, and A Path Forward
This addresses the problem of making general-purpose AI systems trustworthy for industries and users by highlighting unique risks and proposing a framework, but it is incremental as it builds on existing responsible AI principles and techniques.
The paper identifies that modern general-purpose AI systems, such as large language models, pose risks like hallucinations and toxicity due to their high degree of freedom in output, unlike traditional task-specific AI. It proposes C2V2 desiderata (Control, Consistency, Value, Veracity) to guide the development of responsible AI by formally modeling requirements and combining techniques like AI alignment and retrieval-augmented generation.
Modern general-purpose AI systems made using large language and vision models, are capable of performing a range of tasks like writing text articles, generating and debugging codes, querying databases, and translating from one language to another, which has made them quite popular across industries. However, there are risks like hallucinations, toxicity, and stereotypes in their output that make them untrustworthy. We review various risks and vulnerabilities of modern general-purpose AI along eight widely accepted responsible AI (RAI) principles (fairness, privacy, explainability, robustness, safety, truthfulness, governance, and sustainability) and compare how they are non-existent or less severe and easily mitigable in traditional task-specific counterparts. We argue that this is due to the non-deterministically high Degree of Freedom in output (DoFo) of general-purpose AI (unlike the deterministically constant or low DoFo of traditional task-specific AI systems), and there is a need to rethink our approach to RAI for general-purpose AI. Following this, we derive C2V2 (Control, Consistency, Value, Veracity) desiderata to meet the RAI requirements for future general-purpose AI systems, and discuss how recent efforts in AI alignment, retrieval-augmented generation, reasoning enhancements, etc. fare along one or more of the desiderata. We believe that the goal of developing responsible general-purpose AI can be achieved by formally modeling application- or domain-dependent RAI requirements along C2V2 dimensions, and taking a system design approach to suitably combine various techniques to meet the desiderata.