{"id":79944,"date":"2026-07-17T20:04:28","date_gmt":"2026-07-17T20:04:28","guid":{"rendered":"https:\/\/one.sightlinemg.com\/airforcetimes\/uncategorized\/2026\/07\/17\/let-the-ai-do-it-former-air-force-secretary-pits-machine-against-human-operators\/"},"modified":"2026-08-08T17:24:00","modified_gmt":"2026-08-08T17:24:00","slug":"let-the-ai-do-it-former-air-force-secretary-pits-machine-against-human-operators","status":"publish","type":"post","link":"https:\/\/one.sightlinemg.com\/airforcetimes\/industry\/techwatch\/2026\/07\/17\/let-the-ai-do-it-former-air-force-secretary-pits-machine-against-human-operators\/","title":{"rendered":"\u2018Let the AI do it\u2019: Former Air Force secretary pits machine against human operators"},"content":{"rendered":"\n

As the U.S. Air Force moves to increasingly incorporate artificial intelligence<\/a> into operations, the force needs to train human operators effectively to use the tools. But the technology<\/a> might make a better fighter than the human at times, a former Air Force secretary said.<\/p>\n\n\n\n

At a Thursday roundtable discussion hosted by Mitchell Institute for Aerospace Studies, Frank Kendall<\/a>, the 26th Secretary of the Air Force, claimed there are some instances when machines can outperform human operators, since the latter aren\u2019t \u201cperfect.\u201d<\/p>\n\n\n\n

\u201cWhen you get to the point where the AI is doing a better job of deciding who to kill than the person, let the AI do it,\u201d Kendall said during the roundtable. \u201cBut you got to be confident of that.\u201d<\/p>\n\n\n\n

To ensure that AI capabilities<\/a> aren\u2019t outpacing the human operator in the loop, the service needs to verify and test that the tool has been designed to do exactly what it\u2019s meant to, with specific rules it should perform with, Kendall said.<\/p>\n\n\n\n

The former secretary \u2014 who served in the role from 2021 to 2025 during former President Joe Biden\u2019s administration \u2014 highlighted that the service needs to train the human operator to use the tool in the way it was intended and allow them to have many opportunities for practice.<\/p>\n\n\n\n

Intensive simulation can be used to monitor the human operators during continuous machine learning to evaluate how well they are using the tools, what decisions are being made and if the interventions made are necessary or not, he said.<\/p>\n\n\n\n

For example, Kendall said that in an Army range when a rifleman is going against pop-up targets and deciding whether to shoot, \u201cIt is easy to imagine AI shall be perfect at that.\u201d<\/p>\n\n\n\n

\u201cOperators are not perfect, and so I think when you get to that, let the AI do it,\u201d he said.<\/p>\n\n\n\n

Kendall said AI would never be afraid nor tired, unlike a human operator. The machine, he added, needs to start out better than the human, even when the human is at their best.<\/p>\n\n\n\n

\u201cWhen you send armed people out to do something, you run some risk because of human emotions and so forth,\u201d he said.<\/p>\n\n\n\n

The former secretary used the example of a cop pulling over a citizen. He said there\u2019s an \u201cawful lot\u201d of people in America that would rather be stopped by a robot than a cop in traffic because you can program the robot to not shoot at all \u2014 or only if a certain event happens.<\/p>\n\n\n\n

Kendall highlighted that versions of AI have been used in the military for years. He relayed an experience<\/a> during which he boarded an F-16 fighter jet modified with AI. With the touch of a button, the aircraft flew itself, he said. <\/p>\n\n\n\t