Are DL networks understanding shape or only texture patterns?

every week or so I search for any new article online about deep learning, this week bumped into this one:

it includes statements like:

“which shows that the artificial intelligence network looks at the texture of an object more so than its shape,”

"“It’s absolutely reasonable for the golf ball to come up, but alarming that the teapot doesn’t come up anywhere among the choices,” Kellman said. “It’s not picking up shape.”

Humans identify objects primarily from their shape, Kellman said. The researchers suspected the computer networks were using a different method."

That’s very interesting, any opinions from the experts at the community? do you agree that DL networks are not understanding shape and instead are picking up texture patterns only?

thank u for your insights which help us learn more and more :slight_smile: