When a Tesla vehicle hits a pothole in the road, that data is transferred through a network so that the next Tesla, using autopilot controls, doesn’t hit the same hole.

Air Force Gen. Stephen Wilson, the vice chief of staff, said he’d like to see how that can be put into the machines of the military.

Speaking at the New America Foundation’s Future of War conference Monday, Wilson said that if that kind of machine learning can be used for cars, he would like to see how it could be applied to ground vehicles from MRAPs to Humvees.

And then, “Can I do the same thing in the air?”

“Can I do the same thing in space?”

“Then, can I connect space to air to ground?” he said.

“Artificial intelligence and machine-learning are going to be a key part of everything we do,” he said.

More than just avoiding obstacles, AI can help pilots and drivers make decisions and avoid threats as well as target adversaries.

But the focus, he said, will remain on keeping a human in the loop, especially with weapons systems.

Moderator Peter Singer, a New America senior fellow and author of “Ghost Fleet: A Novel of the Next World War,” pushed further, asking the four-star to define how the human operates in the weapons system loop.

In an air combat scenario would a human decide whether to fire on a target? Singer asked Wilson.

Or would it look more like air defense systems in which the machine identifies the target and fires unless the human vetoes the shot?

“I think we’ll have to learn both ways, have to experiment both ways,” Wilson said.