AI Engineering: A National Initiative

Created February 2021

One of the biggest challenges facing the broad adoption of Artificial Intelligence (AI) technologies and systems is knowing that AI systems will work as expected when they are deployed outside of closely controlled test environments. To overcome that challenge, individuals and organizations need to know the right questions to ask and solutions to use when attempting to use AI in real-world settings. AI engineering is an emergent field of research and practice that combines the principles of software engineering, systems, computer science, and human-centered design to enable implementation of AI systems in accordance with human needs for mission outcomes. While much work to date has focused on AI capability development, AI engineering asks what is needed to develop, deploy, and maintain AI systems in unpredictable and chaotic environments.

The SEI is home to a National Initiative aimed at establishing and growing the discipline of AI engineering. The work conducted will lay the groundwork for the practices, processes, and knowledge to build new generations of AI solutions. Alongside partners from defense and national security, industry, and universities, the SEI will guide the creation of a multi-year research and development roadmap and foster development of AI systems that are human-centered, scalable, robust, and secure.

Join Us

A national initiative requires funding, collaboration, and sustained commitment from stakeholders. Only by leveraging our collective resources will we be able to fuel this initiative to unlock the full potential of AI for Defense and National Security and thus achieve the vision of creating viable, trusted, and resilient AI systems. Our network of partners works to source and share research, shape the evolving state of the art, and define gaps and needed capabilities. Join us today!

Collaborators: Conduct and Shape the Research

Conduct research at your organization
Help shape an evolving research and development roadmap
Peer review research
Assist in the aggregation and synthesis of research and results
Provide use cases and data sets
Identify critical AI Engineering opportunity areas

Co-Funders: Fuel the Research and Operations

Funding and in-kind resources are essential to this initiative. Join other funders by providing funding for directed research on activities most aligned with the needs of your organization or provide funding that is used to support broader priorities and operations.

Advocates: Spread the Word

Co-host events, including talks and conferences
Connect us with other stakeholders
Share our research in your organization

AI Engineering National Initiative Projects

The SEI has a rich portfolio of research and expertise to add to the national initiative. The following projects provide a view of that portfolio:

Train, But Verify – This research in robust and secure AI seeks to train machine learning systems to act against more than one threat at a time and verify those systems against realistic threat models.

Inverse Reinforcement Learning – This research in scalable AI uses a type of imitation learning to translate learned information into real-world scenarios.

Knowing When You Don’t Know: Engineering AI Systems in an Uncertain World – This research will improve the robustness, reliability, and maintainability of ML systems by providing benchmark methods for quantifying uncertainty and creating techniques to identify changes in deployment environments that cause uncertainty.

Designing Trustworthy AI for Human-Machine Teaming – This research in human-centered AI guides AI development teams to create AI systems that are accountable, de-risked, respectful, secure, honest, and usable.

Software Engineering Institute