Alexander Mattick says current methods are optimized for impressive demonstrations, while their failure frequency and the consequences of errors—especially in the physical world—are insufficiently clear.

“it up to look very nice in demos and to actually do some really ex impressive stuff but how do you know like how many times it failed to do something and what is the of course the consequence for like juggling balls is really low but what are the consequence of doing something wrong and in like the physical world this is this is a lot bigger of a problem”