I would like to comment, actually. I have a couple of things.
One is that, particularly in AI, actually having gender-diverse people doing AI is really important. We know there are biases built into those systems that are likely, because the viewpoints of the people creating those systems were from a very narrow part of society. In particular for that area, I can understand that.
The other thing, just to reinforce what my colleague has just said, is that there are data that suggest that people are being filtered out early. There's one particularly good study from the University of British Columbia done by a vice-dean there. What they showed was that if you look at the pool of applicants versus who gets on the short list, and versus who gets invited for an interview and gets hired from the short list, racialized people drop out of the bucket, but if they get invited, they get hired.
What they did, then, was actually—