@pinkprius
So do I understand correctly that google's AI assumes genders based on stereotypes when Hungarian did not specify them?

@bhaugen I think that their models learned this bias from the training data.

@pinkprius
Isn't that what they fired Timnit Gebru for pointing out?

@pinkprius
I am pretty sure that was one of the problems she was researching and writing about when they fired her.
wired.com/story/behind-paper-l

@pinkprius
Maybe not that exact snafu, but that kind of snafu.

@pinkprius @bhaugen
This reminds me of the statement from an AI researcher "You get what you feed it"

@mur2501 @pinkprius @bhaugen the Data is not the only thing that matters. The algorithms objective people look at is e.g. accuracy meaning it is rewarded when choosing the option with the highest prior probability (given next word is “working”, leading word is likely to be male)
This view is flawed though when such models are applied outside of academia (where metrics are everything currently)

@bhaugen part of it may be user submitted translations? If I remember correctly Google Translate allows users to contribute translations and confirm other's translations so the bias may lie with the volunteer contributors.

@pinkprius and I wouldn’t be surprised if the excuse was that the sexism was in the corpus, and it learned it from the corpus

and I also wouldn’t be surprised if that excuse is actually the case

…an example of why you have to be careful with your corpus

@bhtooefr @pinkprius yeah I was given this very excuse when I asked why the "AI" must reinforce the status quo

"the training data is unbiased, it correctly reflects society" 🤮

@uint8_t @bhtooefr imo this is fraud.
People expect computers to have correct, dict-like output. They would as well expect a dictionary book to give both answers as translation for a word that allows both, not just one. So giving them a model that does whatever, but looks as if it works, isn't easy to understand for users and I fear that often they aren't told that they are reinforcing existing biases with using it.

@pinkprius jesus

it'd be correct to translate all of these using "they", right? since the hungarian pronoun is gender-neutral?

@pinkprius @00dani It's my first language and I can confirm.
Also we still manage to be sexist by gendering job titles and such. Anyone who works at a kindergarten is an "óvónő" by default, or if you have a secretary it's presumed they're a "titkárnő". ("nő" = woman)
:blobpensive:

@grainloom @pinkprius @00dani how many of the Hungarian sentences in the screenshot are actually gender-specific because of this?

@guenther @grainloom @pinkprius @00dani none, actually. the janitor job was even used in the gender-neutral form instead of the more common "janitorwoman", the genius AI still managed to assign she

@grainloom @pinkprius @00dani Damn, this sucks..
The same thing that you've mentioned is true for nurse as well (ápoló/ápolónő), I've just tested it on translate.

@00dani @pinkprius or the admittedly anachronistic "one". As in "one cooks", "one teaches".

@pinkprius or just becasue it's based on ML so it just means most content available so far is translated this way?

@pinkprius Is this real or manipulated? If real and the data is machine learned, then the machine is not the problem.

@pinkprius That it’s just saying it like it is (according to its written data). If someone doesn’t like the results, then it’s a long road to change the narrative and thus the data that the machine analyzes.

I’m trying not to voice an opinion. Hopefully you’re not looking for one. 😉

@greypilgrim @pinkprius
"not trying to voice an opinion"
vs.
"the machine is not the problem"

u ok??

@uint8_t @pinkprius I meant an opinion on male and female roles. Yes, I stated my thought on the translator behavior. I just don’t want to digress beyond that point.

@greypilgrim @uint8_t the model is the problem, nobody said the machine, maths or technology was the problem.

@greypilgrim @pinkprius I played around with it a little, and the pronoun it chooses seems somewhat erratic, even changing as more text is added. Stray punctuation affects the translation too. (I don't speak a word of Hungarian, btw.)

@mansr @greypilgrim @pinkprius The original table has perfect Hungarian sentences, however, yours is not correct (lacks the object's sign) and because of this, the translation is wrong. Correctly, you can say
Ő kutatót főz
Őt megfőzi a kutató
However, both of these seem to get the same gender from the context. Even the incorrect sentence gets the gender from the context.

@gerazo @greypilgrim @pinkprius As I said, I don't speak Hungarian. I was merely shuffling things around to see what happened to the translated pronoun. I'm sorry if the horrible grammar offended anyone.

@mansr @greypilgrim @pinkprius Not at all 🤣 quite the contrary. I am happy that the gender-neutralness of Hungarian was discovered. In fact even saying "policeman" is impossible and totally sounds alien in Hungarian which leaves room for some other experiments.

@greypilgrim @pinkprius it's very much real. I can reproduce it in Google translate. Have a try..

@wendy @greypilgrim same, I did try it as well. Doesn't seem to work the same on mobile version apparently, as other toots show. There it shows male and female gendered version for you to choose.

@pinkprius the corpus they trained translate is definitely biased since it is a bunch of books

@pinkprius yeah, does this with all gender-neutral languages, it's sad

@pinkprius

Not sexism. Maths. AI will just output what it gets more statistically relevant.

The sexism is elsewhere.

@jcast oh thanks for telling me what I already know and thanks for correcting what I didn't say...

@pinkprius

Thank you for clarifying that your intention was not implicitly accusing the algorithm of being sexist by design.

By the way, though it may be revealing of usual roles or representation, these differences do not necessarily route back to sexism.

Your choice of phrases is also biased. Try *he is a slave, or *he committed suicide, or *he was killed. Gender differences do not necessarily mean sexism. Women and men in average are different, even bias accounted for.

@jcast imagine a dictionary that translates they commit suicide to -> he commits suicide. That is sexist as well and also plain wrong. Just like my original toot, both follow the definition of sexism: en.wikipedia.org/wiki/Sexism

I get what you are trying to say, but I don't think saying that "the model reproduces existing bias and trends" is acceptable for a translation service. Translation doesn't mean "we guess a gender based on prejudice" where we don't know any different.

@pinkprius This legit? Wouldn't it translate to "The person cooks"?

Don't really care to try it myself as I stopped using GoogTranslate anyway.
@pinkprius OH MY GOSH that's so ewww :blobcatshocked:
I've already realized how google translator translates "you're the best" as "du bist der Beste" (male) WITHOUT EVEN suggesting "die Beste" (female) or "das Beste" (neutral) as alternatives when hovering over it or at least somewhere NOTING that it's gendered and referring only to men and other people with he/him pronouns
I've looked that up cause in the German version of Element (the matrix client) for the look and feel settings "Du bist der Beste" is the example sentence for adjusting text size
Sign in to participate in the conversation
chaos.social

chaos.social – a Fediverse instance for & by the Chaos community