Certifiable Artificial Intelligence Through Data Fusion

TL;DR

This paper discusses challenges and proposes methods for certifying AI systems, emphasizing data fusion techniques to establish performance bounds and improve reliability in applications like object recognition.

Contribution

It introduces a framework using data fusion and testing procedures to certify AI performance and reliability, addressing current certification challenges.

Findings

Data fusion can support AI certifiability by providing performance bounds.

Operational testing procedures help manage AI system expectations.

Use case demonstrates improved object recognition reliability.

Abstract

This paper reviews and proposes concerns in adopting, fielding, and maintaining artificial intelligence (AI) systems. While the AI community has made rapid progress, there are challenges in certifying AI systems. Using procedures from design and operational test and evaluation, there are opportunities towards determining performance bounds to manage expectations of intended use. A notional use case is presented with image data fusion to support AI object recognition certifiability considering precision versus distance.