The Pentagon needs high-quality AI data

The goals of the Pentagon for artificial intelligence demand data of a high quality.

Craig Martell, the Pentagon's "AI czar," says that "really high-quality data" must be collected first in order for the U.S. Department of Defense to be able to use and trust AI. He said this at the DODIIS Worldwide Conference in Texas on December 13. He has been in charge of the Chief Digital and Artificial Intelligence Office for eight months, and this task has become his main job.

What are we supposed to do at the CDAO? At first, we thought that our job at the CDAO was to make modelling tools for people in the government. "We don't think that any longer," he said. "We say that we want to build the scaffolding around the model. "We want to help people build that model and use it."

The CDAO was set up in December 2021, but it didn't really get going until June. It took over the jobs of the Joint Artificial Intelligence Center, the Defense Digital Service, the Advana auditing platform, and the chief data officer.

In short, the office is thought to be in charge of and help speed up data analytics and AI at the Pentagon, where they are becoming an increasing focus of spending, testing, and use.

Martell said, "I am sure that when most people say they want AI, what they really want is a good dashboard that just tells them where their stuff is." "So, we think our job is to make sure the data is correct and then give a very strong layer of data analytics."

A report from the Government Accountability Office that came out in February said that the Defense Department was working on more than 685 AI projects as of early 2021. Several of these projects were related to major weapons systems.

AI is powered by a lot of reliable data and training. This digital blood allows AI to help with navigation and recognising targets, as is expected on the Army's Optionally Manned Fighting Vehicle, as well as with predicting maintenance and figuring out how to get supplies to where they need to go. Not all data is the same, though, which makes Martell and his team's job harder.

"First of all, we have a lot of information." He said at the conference, "We have a lot of data spread out all over the world." Some of that data is going to be very useful for making decisions on a large scale. And some parts of this information won't be useful at all for making decisions. So, a big part of our job is to figure out how to handle all of this data so that the data that can help people make decisions is easy to find when they need it.

In April, Martell was named chief digital and AI officer. He took over for John Sherman, who filled the role on an acting basis and is still the chief information officer at the Pentagon. Martell used to work for Dropbox and LinkedIn. At Dropbox, he was in charge of machine learning, and at LinkedIn, he led a number of AI teams and projects.

Martell said at a previous event that he took the government job because of how important it was.

"There aren't many people with a background in both AI and the government," Martell said in June. "So when the deputy secretary of defence calls you and says, "We'd like you to take this job," you have to think really hard about why you wouldn't take the job, not why you would. "And I think it's very important to do this mission right."