Login | March 28, 2024

Federal defense money dedicated for AI development

RICHARD WEINER
Technology for Lawyers

Published: March 3, 2023

The 2022 annual National Defense Authorization Act (NDAA), passed in the last quarter of the year, has money in it for developing artificial intelligence (AI) systems to aid the national defense effort.
Of course, artificial intelligence isn’t really artificial intelligence. I reserve the term “intelligence” for living things, not algorithms. The more accurate term is “machine learning,” and the questions that need to be asked and answered about any given machine learning is how it will be applied to some given situation.
I’ve also seen the term “automated decision making” in the mix (although that could also apply to blockchain activities like smart contracts, or instant online loan approvals, for instance).
In the case of the latest NDAA, the bill creates a pilot program, to be run by the Office of Management and Budget, to develop a path for modernizing the interfacing of various data silos, developing four use cases to show how it could be done.
In addition, the bill requires agencies to look at commercial AI products that can leverage AI and that “(i) operate in secure cloud environments that can deploy rapidly without the need to replace operating systems; and (ii) do not require extensive staff or training to build.”
Sounds like they are anticipating Microsoft putting $10 billion into OpenAI, doesn’t it?
Well, no, they couldn’t have done that. That wasn’t announced until months later.
The pilot program is going to prioritize use cases where AI can drive “agency productivity in predictive supply chain and logistics,” like predictive food demand and optimized supply, or predictive medical supplies and equipment demand, or predictive logistics for disaster recovery, preparedness, and response (very useful in hurricane season, for instance).
There were numerous bipartisan attempts to pass data privacy and other cybersecurity bills in the last (lame duck) Congress, but they mostly failed.
One effort that did pass was the requirement that government cloud computing contractors have to be certified by a process called the Federal Risk and Authorization Program (FedRAMP).
This includes analyses by Covington & Burlington LLP. Thanks to them.


[Back]