A challenge in defense acquisition can be the ability to keep up with the speed of innovation in emerging technology, but specifically, the acquisition of Artificial Intelligence (AI) systems can pose a unique challenge, especially if the technology comes from relatively new start-up companies.
Often these companies can be at the leading edge of technology. So a team of researchers at the Naval Postgraduate School (NPS) have focused their efforts to develop a set of tools to support the acquisition of organically-developed AI systems.
NPS’ Information Science Professor of Research, Dr. Johnathan Mun, who teaches quantitative research methodologies and decision analytics, says that acquisition research is a key factor in keeping the United States ahead in the era of Great Power Competition. Mun notes that companies like Lockheed Martin or Northrop Grumman have a proven track record with the Department of Defense (DOD) in making high-quality products, however, the acquisition of AI technology by start-ups makes things a little more complex.
“Major innovations have happened before with two guys in a garage coming up with some really cool, potentially game changing, revolutionary, and disruptive technology,” said Mun. “There is a huge potential for high-risk failures in this field because this is brand new technology. When a couple of really smart people in a garage somewhere invent a new algorithm we have to determine if it’s worth our time and money.”
When the DOD discovers these new start-up companies and products, they have to rely on acquisition research to determine the risks of pursuing the company.
“So, how do you analyze AI?” asks Mun.
Mun’s team, comprised of three other NPS researchers, have provided an answer to that question. By using already-developed equations in concert, they can help determine the potential worth of technology emerging from start-up companies around the world.
“In the analysis, we need to apply 80 percent advanced decision analytics approaches and 20 percent qualitative intangible factors,” said Mun. “80 percent is math because no matter what, one plus one equals two. The other 20 percent depends on things like what timeline we are working on or how confident we are that the technology will make our jobs easier or more capable. We also look at how would this technology hurt us if it got into the wrong hands, or if we could add it to an already existing program.”
Defense acquisition actually requires input from multiple different sources. The warfighter has a need, the decision makers have a budget, and the engineers have to be able to address those needs and then the DOD has to acquire those solutions.
“Acquisitions is a very complicated topic,” said Mun. “You need to have different experts from multiple fields.” Mun added that it all comes down to a cost benefit analysis and understanding the risk associated with the acquisition of technology, and understanding its long term value.
According to one of the team’s other researchers, NPS Information Science Professor Dr. Tom Housel, another thing that sets these start-up companies apart is the pace at which they are being created and the pace at which they are innovating. He says that this new “industrial revolution” is moving at a very fast pace.
“We have to be able to screen through these new technologies at a fast and accurate pace,” said Housel. “If you show me a model and tell me what it does, then we can tell you how much value it adds.”
Mun noted that the interdisciplinary aspect of acquisition is what makes NPS the perfect place for this research.
“We have these great professors who have experience in AI and information technology and acquisition,” said Mun. “But we also have these great students who are the warfighters and will ultimately be using the product. We are able to collaborate together to determine the best path forward.”
When it comes to acquisition research processes, there is no standard way to assess every situation.
Mun noted, “At the end of the day, our job is to provide actionable intelligence to the senior leadership and decision makers to help make the acquisition process easier.”
| Date Taken: | 06.10.2021 |
| Date Posted: | 06.10.2021 17:26 |
| Story ID: | 398639 |
| Location: | MONTEREY, CALIFORNIA, US |
| Web Views: | 94 |
| Downloads: | 0 |
This work, Evaluating Artificial Intelligence: NPS Researchers Seek to Improve Acquisition of AI Systems, by PO2 James Norket, identified by DVIDS, must comply with the restrictions shown on https://www.dvidshub.net/about/copyright.
