Comment by upwardbound | Hacker News Reader

Comment by upwardbound | Hacker News Reader

Parent Full thread

If Apple copies anything from us, I hope they copy this:

The computer-generated voice we used in the Spectacles V4 out-of-box experience tutorial is a gender-neutral voice. We did this as the start of a journey of attempting to convince the software industry to stop making A.I. assistants be female so often. Having female A.I. assistants perpetuates the gender wage gap by creating the idea that assistant jobs are a women's role. This is a big issue that needs to be tackled and I'm proud to have helped with it in a small way. If anyone reading this works on Siri, Alexa, Cortana, or Google Assistant, please consider asking your management to adopt gender-neutral voice technology.

Here is UNESCO imploring companies to do this: https://en.unesco.org/EQUALS/voice-assistants

"[UNESCO] advises companies and governments to, among other actions: end the practice of making digital assistants female by default; explore the feasibility of developing a neutral machine gender for voice assistants that is neither male nor female"

This is such a weird response to a technical criticism, do you have anything to say about anything the parent poster said?

Oh, I totally understand the parent poster's perspective and they might 100% be right. If I were a betting man I would bet on Apple; who wouldn't? Only time will tell; I wasn't trying to refute the parent poster in any way. Was just highlighting an aspect of Specs that I really believe in that isn't in the press and wanted to put this somewhere that someone from Apple might hopefully read it.

Perhaps you didn’t realize Apple already made exactly this change in a recent update?

correct, I totally did not and that is amazing!! Just three more to go :) Alexa, Cortana, and Google Assistant.

Someday consumer grade mobile voice assistants will be feature complete and bulletproof enough that adding a gender-neutral voice will be the highest user value feature to invest limited development resources into.

Based on my recent experience trying to support my elderly mother's desire to "talk to her phone to do things", that day is not imminent.

Alternatively if the voice were male you'd have people complaining that it reinforces the idea that technology is male centric and it further alienates women in tech.

But yeah, I'd love gender neutral voice assistants as well.

iOS offers the choice between male and female assistant voices and has removed a default.

https://www.theverge.com/22364180/how-to-change-siri-voice-i...

That's amazing!!! I didn't know they had implemented this now, thank you so much for sharing!

>> Having female A.I. assistants perpetuates the gender wage gap by creating the idea that assistant jobs are a women's role.

is there any proof of this or does it just sound true therefore its true? Specifically theres two claims here that i don't see any data backing up:

1. Female voices of AI creates the idea that assistants jobs are a women's role.

2. Assistants jobs being perceived as a woman's role somehow makes the (nonexistent) wage gap worse. Despite the fact that assistants tend to make more money than garbagemen. (a field dominated by men)

There is human factors work on the gender of voice interfaces. A notable real world example is the voice that firmly proposes 'Pull up!' if you're about to fly into the ground in many aircraft.

Wikipedia is missing refs, but here's the entry on choosing the gender of this voice

https://en.wikipedia.org/wiki/Voice_warning_system

[edit] originally quoted the first part of the section that cited early work finding women's voices to be more effective, but later work challenges this finding. I removed the quote to avoid implying I believe either way.

thats interesting but nonetheless has no bearing on these specific claims. if anything it counters the claims as it says a female voice is more authoritative whereas the original poster said it makes them seem like an assistant or submissive and somehow affects their wages.

also your example is one where they shaped the assistant based on the humans perception. whereas the poster is claiming the assistant shapes the humans perception.

Both of these topics are covered extensively in the long-form version of the UNESCO report. The link I posted was just a summary; here is the long form: https://unesdoc.unesco.org/ark:/48223/pf0000367416.page=85

that doesnt really have any real sources for the two assertions. it references this:

https://science.sciencemag.org/content/356/6334/183/

which doesnt really have anything to do with the point. Just vaguely related to biases inferred from NLP. It also just quotes a few sociology professors who themselves dont provide proof behind their assertions. And then also points to:

https://wappp.hks.harvard.edu/publications/psychology-implic...

which itself just talks about biases in general and doesnt have anything to do with the voices of AI assistants.

If you're interested you can also look at more sources in the first two pages of this Google Scholar search: https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=voic...

By the way, what makes you say the gender wage gap isn't real? To many people that comes across as if you were denying climate change, and decreases your credibility overall. Just trying to engage as best I can.

because the widely quoted figure of 79 cents to a dollar is only if you blindly compare incomes without adjusting for hours, experience, education, or even role.

obviously a part time elementary school teacher with two years experience will make less than a full time investmenet banker that works 80 hours a week and has 10 years experience.

Once you adjust for all the relevant factors the effect completely disappears and there isnt a gap. (with some studies find maybe 1-2% within the margin of error).

so at the very least its very misleading to say they get paid less for the same role and same expereince.

if you want to go into why they may go into different fields or work less hours and how that reflects on society thats fine.

But its grossly misleading to take those nunmbers and blindly compare them and say its "sexism". Actually indian and asian woman make more than white men. So white men must systemically be oppressed. Or maybe they choose different jobs. (i still remember when google had to adjust their L4 software engineer pay to raise mens pay, because women actually made more!)

those things you're adjusting for are part of the problem! If you're going to adjust away all possible explanations of the problem, then of course you won't see the problem anymore!

not really. thats how you isolate a variable. people are postulating the reason for the gap is sexism. so you isolate all the variables until the only difference is sex. and then you see if theres a difference. its basic statistical methodology. in fact alot of people are mistakenly under the impression that women make less for the same role/experience level, which my statistic shows isnt true.

if you dont adjust for something then its just an idiotic comparison. like i said, non-adjusted asian women make more than white men. (so do nigerians) But you arent crying about asian women supremacy. or the the system is racist/sexist against white men.

now of course you can argue "sexism of the gaps". that the reason women choose jobs that make less is sexism itself. but youd have to prove that assertion, and not simply that hey prefer jobs that tend to make less like social work and teaching.

and youd have to explain why the so called wage gap is bigger in the most gender-equal countries like norway and denmark whereas its the smallest in the least equal countries like in the middle east.

Reply on news.ycombinator.com