"

The models encode many biases and stereotypes.

    Well, sure they do. They model observed human's language, and we humans are terrible beings, we are biased and are constantly stereotyping. This means we need to be careful when applying these models to real-world tasks, but it doesn't make them less valid, useful or interesting from a scientiic perspective."

Not sure how this can be seen as dismissive.

>Yoav can dismiss this because it just doesn't affect him much.

Maybe just maybe someone named Yoav Goldberg might maybe be in a group where bias affects him quite strongly.

In fact, the author even says this argument is "true but uninspiring / irrelevant". He's just deciding not to focus on that aspect in this article.

Or maybe he is blind to or unaffected by such biases either due to luck or wealth or other outliers. Especially as a Jewish person in Israel. There are always plenty of people in minority groups that feel (either correctly or incorrectly) that bias doesn't affect them. Take Clarence Thomas for example, or Candace Owens. Simply being a member of a minority group does not make your opinion correct. Thomas even said in recent oral arguments that there wasn't much diversity in his university when he attended and so he doesn't really see how a diverse student body is beneficial to one's education.

Or maybe he recognizes that it’s literally impossible to train a system to output a result that isn’t biased. Creating a model will result in bias, even if that model is a calculator. You put your perspective in its creation, its utility, its fundamental language(s) (including design language), its method of interaction, its existence and place in the world. If you train a model on the web, it’ll be billions of biases included, including the choice to train on the web. If you train on a “sanctioned list,” what you include or don’t include will also be bias. Even training on just Nature papers would give you a gigantic amount of bias.

This is what I really don’t like about the AI ethics critics (of the woke variety): it’s super easy to be dismissive, but it’s crazy hard to do anything that moves the world. If you move the world, some people will naturally be happy and others angry. Even creating a super “balanced” dataset will piss off those who want an imbalanced world!

No opinion is “correct” - they’re just opinions, including mine right now!

>Simply being a member of a minority group does not make your opinion correct.

Nor does being a member of the majority make yours incorrect.

I mean he is presumably Jewish and lives in Israel, so I would guess he knows quite a bit about being a minority and experiencing bias.

Jews are a minority in Israel?

No of course not, nor did I claim they are.

Similar question to one above, I don't follow why the (positive) ad hominem bolsters or detract from the arguments.

Because the claim of the comment I was responding to was that Yoav doesn't experience bias and consequently dismisses it, so in this case my comment is a direct refutation of the argument.

We're the "majority" by virtue of internationally gerrymandered borders. In the region as a whole? Yes, we're an indigenous minority.

The Boers are an indigenous minority in southern Africa, but in the 80s I wouldn't have used the Boers as an example of people who really understand the experience of bias as a minority.

>The Boers are an indigenous minority in southern Africa,

No they're not.

What does the word "indigenous" mean?

That's true, he probably sees a lot of anti-Palestinian bias on a regular basis.

And globally a disproportionate amount of anti-Israeli and anti-Jewish bias, obviously.

Inside Israel?

>Yoav can dismiss this because it just doesn't affect him much.

I'm sure nobody named Yoav Goldberg has ever had a literal war waged on him for reasons of petty identitarian bias /s.

Don't understand the downvotes but I do disagree. What the industry needs is not overtly racist hiring practises but rather people who are aware of these issues and have the know-how and the power to address them.

I'll take an example. I'm making an adventure/strategy game that is set in the 90s Finland. We had a lot of Somali refugees coming from the Soviet Union back then and to reflect that I've created a female Somali character who is unable to find employment due to the racist attitudes of the time.

I'm using DALL-E 2 to create some template graphics for the game and using the prompt "somali middle aged female pixel art with hijab" produces some real monstrosities https://imgur.com/a/1o2CEi9 whereas "nordic female middle age minister short dark hair pixel art portrait pixelated smiling glasses" produces exclusively decent results https://imgur.com/a/ag2ifqi .

I'm an extremely privileged white, middle-aged, straight cis male and I'm able to point out a problem. Of course I'm not against hiring minorities, just saying that you don't need to belong to any minority group to spot the biases.

Our biases reflect statistical relationships that people observe. There is a body of evidence that suggest that these biases are rather accurate. I don't see why we would want to remove them from our models.