Also: what does "we recorded many participants performing tasks involving deceptive walking" mean? Does this AI detect those who perform deceptive walking, or people who are actually deceptive walking?
Also: what does "we recorded many participants performing tasks involving deceptive walking" mean? Does this AI detect those who perform deceptive walking, or people who are actually deceptive walking?
So yeah! No surprise! They told the participant to act sketchy, and they did! What did the participant do? They put their hands in their pocket and looked around a bunch more than normally. Why? Because they were basically told to! That's not "deceptive walking".
Where is the control for picking up the papers off the floor, or the money off the table? How do we know it's not just that people will walk differently when they are about to crouch down? Maybe that would be an interesting paper by itself, but instead the authors think that this type of research is capable of identifying "deception" with 93% accuracy? Absurd!
There is so much wrong with the experiment, that you could write a rebuttal of almost every decision they made.
This is not deceptive walking, this is honest walking. They are honestly trying to broadcast the signal that they were told to broadcast.
The "deceptive" part is just a frosting layer of prejudice they laid on top.