Something I haven't seen anyone bring up yet:
The author works for Google's "Responsible AI" division. I would very strongly hope that religious people would be welcome there, obviously along with atheists and people or all origins or backgrounds. AIs will soon have very real power over average people's lives. Humanity needs to make sure we get it right.
His 2 main points are:
- religious people being treated without respect or osctracized
- algorithmic bias against religion in Google products
As with all these past discussions (see Timnit Gebru), it's hard to know whether he's really being treated unfairly unless you somehow know him in person. Some people have personality disorders that make them insufferable to work with, religious or not. Personally he seems well-adjusted; it doesn't look like he was trying to proselitize or evangelize; he brought up religious discrimination in passing, was asked to explain what he meant, and privately explained. I don't know why so many here are assuming he's walking around Google asking people if they've heard of "our Lord and Saviour".
>I mentioned that I stood against all sorts of discrimination at Google and listed religious discrimination as one of them. He asked for examples so I privately messaged him with some and he began trying to “educate” me about how those weren’t religious discrimination; they were just signs that I was suffering from “victim syndrome”.
The second issue he brings up (that he says he needs to discuss with a lawyer before he can give concrete examples) also seems plausible enough. Google admits to downranking independent news organisations, politically non-mainstream outfits (socialists[0], subsets of social justice activists, nationalists); given how atheistic most Google employees are, it's plausible that the same would apply to religion. There is also a documented history of religious bias in Google Assistant[1].
If what he's claiming is true, it's indefensible. Algorithms shouldn't have an anti-religious bias. Think: algorithms for hiring, parole/sentencing, or personality testing. Religious discimination isn't a hypothetical, there are people being killed or discriminated against today in various countries because they belong to minority faiths. AIs shouldn't participate in that.
As an interesting tidbit: here's what Marget Mitchell (co-founded Microsoft Research's Cognition group, Google Research's ML Fairness group, fired from Google shortly after Timnit Gebru) said about the author (Lemoine):
> When new people would join Google who were interested in ethics, Mitchell used to introduce them to Lemoine. “I’d say, ‘You should talk to Blake because he’s Google’s conscience,’ ” said Mitchell, who compared Lemoine to Jiminy Cricket. “Of everyone at Google, he had the heart and soul of doing the right thing.”
[0]: https://www.michigandaily.com/viewpoints/op-ed-why-google-ce...
[1]: https://www.newsweek.com/google-censoring-jesus-knows-buddha...