Priority Quality Attributes for Engineering AI-enabled Systems

Abstract

AI systems are built of software. Engineering an AI-enabled system poses some challenges that are distinct from "conventional software". AI-enabled systems are not a monolith - e.g. neural network methods vs. regression based methods. The interaction between software and data touches all of the challenges and architecture considerations we will discuss. We need new methods and architecture solutions to design AI-enabled systems that can be confidently deployed in public sector context.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 26, 2019
Accession Number
AD1085207

Entities

People

  • Ipek Ozkaya
  • Lena Pons

Organizations

  • Carnegie Mellon University

Tags

DTIC Thesaurus Topics

  • Department Of Defense
  • Engineering
  • Guarantees
  • Materials
  • Universities

Fields of Study

  • Computer science
  • Engineering

Readers

  • Distributed Systems and Data Platform Development
  • Software Engineering.
  • Unmanned Aerial System (UAS) Autonomous Capabilities and Mission Reconnaissance.

Technology Areas

  • AI & ML
  • AI & ML - DoD AI Strategy
  • AI & ML - Neural Networks