Published: March 15, 2022
Citation: Computer (IEEE Computer) vol. 55, no. 3, (March 2022) pp. 82-86
Author(s)
Laura Freeman (Virginia Tech), Feras Batarseh (Virginia Tech), Richard Kuhn (NIST), M S Raunak (NIST), Raghu Kacker (NIST)
Widescale adoption of intelligent algorithms requires that Artificial Intelligence (AI) engineers provide assurances that an algorithm will perform as intended. Providing such guarantees involves quantifying capabilities and the associated risks across multiple dimensions including: data quality, algorithm performance, statistical considerations, trustworthiness, security, as well as explainability. In this article we discuss the formalization of important aspects of AI assurance, including its key components.
Widescale adoption of intelligent algorithms requires that Artificial Intelligence (AI) engineers provide assurances that an algorithm will perform as intended. Providing such guarantees involves quantifying capabilities and the associated risks across multiple dimensions including: data quality,...
See full abstract
Widescale adoption of intelligent algorithms requires that Artificial Intelligence (AI) engineers provide assurances that an algorithm will perform as intended. Providing such guarantees involves quantifying capabilities and the associated risks across multiple dimensions including: data quality, algorithm performance, statistical considerations, trustworthiness, security, as well as explainability. In this article we discuss the formalization of important aspects of AI assurance, including its key components.
Hide full abstract
Keywords
artificial intelligence; data quality; software testing; statistics
Control Families
None selected
