Create a free starter account, or log in.

Get access to four free articles per month, as well as our Decisive Edge Newsletters and Daily News Bulletin.

Email Address

Password

Forgot your password? Reset by clicking here

Create a Password

[[ req.met ? '✓' : '•' ]] [[ req.text ]]

Re-enter Password

Passwords do not match

First Name

Last Name

Country

Confirm your account

Please enter the four-digit pin below that has been emailed to you.

Didn't receive the pin yet?

This pin expires in 20 minutes.

[[ error ]]

By creating an account, you agree to the Terms of Service and Privacy Policy.

Stay informed with Shephard's award-winning journalism. Unlock unlimited access and in-depth analysis with a corporate or individual subscription.

Explore your options here

I/ITSEC 2021: Military must address AI disconnect from lab to frontline

30th November 2021 - 11:28 GMT | by David Isby in Orlando

RSS

The Western Air Defense Sector rolled out an AI-based system called Battle Management Training NEXT in late August 2021 at Joint Base Lewis-McChord, Washington. (Photo: USAF/Maj Kimberly Burke)

A roadmap, including a plan to address training requirements, is needed to ensure that US military AI programmes are not lost in translation from theory into practice.

Challenges in integrating AI with deployable, operational US military capabilities were addressed in a tutorial from industry and government experts on 29 November during the opening day of the annual I/ITSEC meeting in Orlando, Florida.

Courtney Crosby, chief AI solutions architect at Booz Allan Hamilton, noted that ‘anywhere from 50 percent to 85 percent’ of all AI programmes fail in combat zones even when their underlying algorithms were technically successful in testing or in laboratory environments.

She attributed this to treating AI as a technical problem, often overshadowing the need to ‘operationalise’ AI as part of an existing mission.

Already have an account? Log in