I could talk about the stats, for sure, and then I'll talk about what we can seek as Mr. Seeback suggested.
You're right. The refusal rates have substantially gone up. I just disaggregated francophone Africa, and per my count, basically what we're seeing is that there's a 27%, essentially, approval rate coming in from francophone Africa from 2016 to 2020. That is in contrast to the main source countries for Canada, which are at around 70%. That division, that gap, is too big.
As to what we're looking for in terms of accountability and what Chinook is, first of all, unfortunately, I'm very sad to report that we actually don't know enough to ask the good questions. However, we do have some indications on what AI good governance could look like when we start asking questions about Chinook.
For example, during the conference at AQAADI in Montreal back in November, Richard Kurland, another immigration lawyer from Vancouver, was suggesting that there needs to be some sort of backstop, at the very least even internally at IRCC, to have some accountability for how these algorithms actually work. What's the assessment?
I'd like to make a contrast. There are two programs that I mentioned in my oral submissions: the first is Chinook; the second one is the artificial intelligence-driven advanced data analytics, which Will can talk about a little more. The latter, advanced data analytics, actually has some QA backing it, whereas with Chinook, we have absolutely no idea.
We've tried for the last 60 days to do some ATIP requests and what we're getting is a lot of push-back. We keep getting extension delays of about 180 days or even a year on our requests, essentially trying to push the issue forward. Therefore, we can't really comment as to what variables we need.
Perhaps the committee can actually get that data—