Do you think affirmative action is a good and fair way to increase diversity in tech?
1 Like
This mean that I’ll never find a job! lol
1 Like
I think Fast AI forums is a good place to get diverse friends and acquaintance.
5 Likes
How do you think about our opportunity to correct biases in artificial systems, versus the behaviors we see in humans?
(E.g. a sentencing algorithm that can be monitored and adjusted, versus a specific biased judge who remains in their role for an extended period?)
5 Likes
IMHO the absence of a “moral stance” is a moral stance.
1 Like
I would much rather go before a biased Judge than a biased algorithm.
At least there is a possibility that a biased judge can have his mind changed.
Cheers mrfabulous1 
I personally think that it’s good, but it’s not marketed properly. It should be seen not as a program to displace one race over another, but something that is needed to increase diversity of backgrounds, which at the end creates a better and more productive environment.
In response to that, Facebook released a Burmese translation model 
One way I like to think about “fixing bias” is the fact that AI assumes the present is wholly represented by the past when in reality it may not be. Any bias we correct will still be fixing the past, not the emerging present.
6 Likes
Is there an Internal Review Board (IRB) specifically for Tech and AI like there is in public health?
2 Likes
Most governments do rely on companies for these technologies.
If more people and even grassroots campaigns call out these companies about
their technologies and how they are used…
Would developing an effective grassroots campaign on AI ethics along the lines of climate change activism be effective?
1 Like
Is it “correct bias”, “adjusted bias”, or “weighted-by-people bias”? I’m still not sure how to call this.
Anyone know of a research paper or practical application where a generative model was used to overcome bias? E.G lack of females of color in dataset
1 Like
Will rachel’s full course on ethics be available at some point?
5 Likes
That’s a nice font in the presentation, can anyone tell me which is that?
If so, would we be talking about the “great unbiased bias” as a moving target that would require never-ending dynamic approximation?
I suspect just like climate change, multiple simultaneous efforts like grassroots campaigns, tech activism, regulation etc. will be required to make ethics a priority within tech companies.
Grassroots campaigns can certainly get companies to care about reputational risk or financial impact. Though I suspect the cycles of bad PR due to an exposed or potential ethics issue are short-lived or have little to no dent on revenue, so companies fail to fix the root cause which is a lot harder to fix!
1 Like
Car safety is good analogy. When Robert McNamara was head of Ford in the 1950’s, he crunched the data and realized that cars needed more safety measures as they went faster on the new interstate highways. But customers didn’t want safety, sales plummeted, and car companies concluded that safety features would kill sales.
What a great class on an important topic. Thanks Rachel!
1 Like
Thanks!