Live data from Hacker News

Google’s AI-powered ‘inclusive warnings’ feature is very broken

vice.com

961–970 of 1001 posts

Re: Google’s AI-powered ‘inclusive warnings’ feature is very broken

#961

Earlier quoted context omitted.

> words we've been using for decades to describe a concept. Yeah, and so was calling black men "boy" or using the n-word like 50 years ago. Language changes because people realize there are better words that more people are okay with.

Language also adapts over decades so that words are computer/IT terms & have no other connotations. Like most of the stuff in this thread. Have a think about that.

Words do have connotations, that's the reason why people use them - because they refer to similar concepts, even in different contexts.

Re: Google’s AI-powered ‘inclusive warnings’ feature is very broken

#962

Earlier quoted context omitted.

Grammarly is also an OS-wide keylogger. https://www.kolide.com/blog/is-grammarly-a-keylogger-what-ca... They claim not to be a keylogger, but, you know, Amazon claims that Alexa isn't always listening too and/but/yet also that it'll wake up immediately when you say the right wake word. So, I'd take their claims with a lot of salt.

> Amazon claims that Alexa isn't always listening too and/but/yet also that it'll wake up immediately when you say the right wake word Strictly Alexa, the device, is always listening in order to trigger on the wake word. It's just that audio doesn't get sent to Amazon until it hears a wake word (then it just streams the audio to Amazon to process it. locally it's only smart enough to listen for a wake word). This is…

Grammarly really is a keylogger though. In a "plain text being sent to a remote server" kind of way.

Re: Google’s AI-powered ‘inclusive warnings’ feature is very broken

#963

Earlier quoted context omitted.

Many folks I know use "they". Works well enough, I think.

Shall we write "they is pretty" or shall we wonder how many people we are talking about?

We shall wonder about the how many people. This is called the "singular they" and it has been a feature of English for hundreds of years.

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

Re: Google’s AI-powered ‘inclusive warnings’ feature is very broken

#964
post #766

So much of this technology perpetuates the biases they're intended to fix. We put some of our job postings through one of these, and it flagged too many 'masculine' words, like 'analyze', 'leadership', and 'data'. Sorry women, time get back in the kitchen and nurture.

Got a source for that? Or an/the online tool you used? I'd love to try it

Found a hilarious blog entry regarding OP's comment. "Analyze", for one, is unnecessarily masculine in job postings, they say, and then present their tool to alleviate the problem: the Text Analyzer.

Or maybe they want to help people write specifically masculine job postings?

https://blog.ongig.com/writing-job-descriptions/top-10-mascu...

Re: Google’s AI-powered ‘inclusive warnings’ feature is very broken

#965
post #651

Earlier quoted context omitted.

I think you're looking for the word parent?

A female parent is a mother. That's what their children call them, not "parent", with allowances for foster children and older adopted kids calling them by their first name.

I don't think calling a parent 'mother' is something children do by themselves. They call them however they end up calling them, or whatever their parents teach them. Also certainly not 'mother', rather 'mum', 'mom'... and the Turkish children say anne.

Re: Google’s AI-powered ‘inclusive warnings’ feature is very broken

#966

> Emily Lipstein typed “Motherboard” (as in, the name of this website) into a document and Google popped up to tell her she was being insensitive: “Inclusive warning. Some of these words may not be inclusive to all readers. Consider using different words.” It’s amazing how a Fortune 500 company can release a product that chastises you for using the word “motherboard” because it has “mother” in it, while other people…

Birthingpersonboard you bigot

It's actually main board or logic board, as per, for example, the Apple guidelines.

Re: Google’s AI-powered ‘inclusive warnings’ feature is very broken

#967
post #190
post #98

Earlier quoted context omitted.

Not only that, but there are more than three times as many slaves TODAY as ever were in the transatlantic slave trade that is the only one the US knows exists.

There are more slaves added each year in the US today than at the peak of transatlantic trade: https://www.britannica.com/summary/Transatlantic-Slave-Trade... Peak transatlantic: 78,000 new slaves per year. https://thecurrentmsu.com/2021/01/24/prison-labor-americas-s... Current US forced prison labor population: 1-2.1 million.

Wow, lots of replies are trying to downplay the existence ongoing human suffering.

A few things to consider:

- A disproportionate number of prisoners are African American, and that subset are mostly in prison for crimes that whites simply don't go to prison for (e.g., pot, or coke vs. crack).

- US prisoners aren't given nutritionally adequate food for free. They have to buy that at inflated prices from the prison. Once they blow through any savings, their only option (other than malnutrition) is to work for wages that would be illegally low outside of prison.

- As during the transatlantic trade, it is straightforward to buy your way out of forced prison labor.

- A growing fraction of prisoners consider the current system to be eugenics as well as slavery. The prisons take the majority of men while they are at prime child rearing age. It's well known that any that do manage to have kids will probably be forced into a situation where they can't adequately care for those children (dad goes to prison for driving while black or whatever), making the next generation easy targets for the same scheme. (This is coming from current prisoners, not me.)

The other pushback against my comment boils down to "it's legal, so we don't use that word". Similar arguments were made in defense of the transatlantic trade. For instance, black Africans were generally the ones selling slaves, and often used real or imagined criminal records as an excuse. Any objective comparison of the two systems would find that the transatlantic slave trade was roughly as immoral California's three strike law is in practice. You could argue the slave trade was worse because the kids were automatically slaves. In the current system, they're only probably going to prison, but the prisons also aren't paying to raise them.

I suggest reading Things Fall Apart if you're unfamiliar with the African side of the slave trade.

Re: Google’s AI-powered ‘inclusive warnings’ feature is very broken

#969
post #909

Earlier quoted context omitted.

Just as a total aside - why do you think they don't want people using google as a verb? It's not as if it killed Xerox or TiVo.

I think it has something to do with trademarks being lost once a name becomes too synonymous with a product category. e.g. Aspirin, Hoover

Ah! Good answer. Hadn't thought of that.

Re: Google’s AI-powered ‘inclusive warnings’ feature is very broken

#970

This link in the article is insanely, crazily, madly, insultingly sickening to anyone who's not a slave. As I read it I felt like it was some kind of parody of Orwell. Disgusting to tell someone what verbiage they use sincerely and in technical terms is now verboten because it's harmful. To whom? Crazy people? https://developers.google.com/style/inclusive-documentation

I don't think it's the worst thing to encourage people too be thoughtful about their word choice. Is it really "crazy", or are you selling an emotional reaction by personifying your topic as a mentally deranged person? I actually think steering clear of this language is helpful in defusing the outrage culture that is so harmful to our online discourse.

I chose my wording carefully to illustrate that the conveyance of emotion, even through jarring phrases, is an important part of language. You can absolutely disagree with the rhetorical style, but the point is there's a vast difference between your rebuttal to it and an ML mechanism built into your writing program that dissuades you from writing certain types of prose, or persuades you to use language that an opaque AI decided is less triggering.

We don't have to all be helpful, logical or anodyne all the time, either. Repressing what you want to say is unhealthy and also leads to dishonest debate. We don't need to be coddled and protected from some speech by algorithms. What we need to do is learn to see the rhetorical flaws and rebut them - as you just have (although, as I said, those flaws were there on purpose, as a layer of meta-meaning in sarcastic word choice that would fly over the head of an algorithm).

Post reply on HN