DHS wants AI forces working more as a department-wide team

The U.S. government has published three strategic security goals for artificial intelligence and machine learning — both defending against them and using them to protect the nation.

The plan was written by the Science and Technology Directorate (S&T) of the Homeland Security Department. Implicit in the document is the deployment of biometric systems.

An implementation plan is forthcoming, according to the directorate.

Chief among the strategy’s goals is minimizing stovepipe thinking and development within Homeland Security.

One goal calls for developing new AI and machine learning technologies that can be used across the department. A second prioritizes interdisciplinary training for government workers.

A third goal, naturally, calls for using the resulting products to secure the nation.

While actions speak louder than words, the directorate has ordered research and development that advances trustworthy AI as part of the strategic plan. Aims to deliver on this priority include delivering explainable AI, privacy protection and countering bias and adversarial machine learning.

That might be the biggest hurdle for Homeland Security. Even mundane but necessary information about AI development can get redacted.

It is interesting to note that the department wants to develop systems that do not require advanced knowledge or experience in order to use them. Monastic development and deployment are common with technology and often lead to limited and brittle results.

Encouraging as the plan is in many respects, it describes a department-wide stovepipe for development and deployment. There are no mentions of working with the many other related efforts active or soon to be launched across the federal and state governments, or with like agencies in foreign governments allied with the United States.

Article Topics

AI  |  biometrics  |  computer vision  |  DHS  |  machine learning  |  research and development  |  U.S. Government