Toward Verified Artificial Intelligence

Sanjit A. Seshia, Dorsa Sadigh, and S. Shankar Sastry. Toward Verified Artificial Intelligence. Communications of the ACM, 65(7):46–55, 2022.

Download

[HTML] 

Abstract

Verified artificial intelligence (AI) is the goal of designing AI-based systems that have strong, ideally provable, assurances of correctness with respect to mathematically-specified requirements. This paper considers Verified AI from a formal methods perspective. We describe five challenges for achieving Verified AI, and fifteen corresponding principles for addressing these challenges.

BibTeX

@article{seshia-cacm22a,
  author    = {Sanjit A. Seshia and
               Dorsa Sadigh and
               S. Shankar Sastry},
  title     = {Toward Verified Artificial Intelligence},
  journal   = {Communications of the {ACM}},
  volume    = {65},
  number    = {7},
  pages     = {46--55},
  year      = {2022},
  abstract = {Verified artificial intelligence (AI) is the goal of designing AI-based systems that have strong, ideally provable, assurances of correctness with respect to mathematically-specified requirements.  This paper considers Verified AI from a formal methods perspective.  We describe five challenges for achieving Verified AI, and fifteen corresponding principles for addressing these challenges.},
}