[m101]

is this true because training companies have not been training AI for both performance and brevity (or some other metric like that)? If this becomes a much more serious issue surely they would adjust the training processes