Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

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

vice.com

Post reply on HN
Print view View original

901–910 of 1001 posts

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

#901

Post by cuteboy19 »

chc wrote: post #890
creakingstairs wrote: post #866

Earlier quoted context omitted.

The biggest annoyance for me is that the word/woke policing throws out the most important thing: context. We don’t punish people for crimes until we’ve had a look at the context and establish their specific intention (Strict liability withstanding) so why are we not doing the same with words and sentences? Master/slave debacle is a prime example. Yes it can be used offensively. But for gods sake, look at the context!…

That seems like an odd example. The "specific intention" of the word "slave" in computing is unquestionably to draw an analogy to slavery, so if somebody objects to the word on the grounds that it references an actual atrocity, they're factually correct. I think what you're trying to say is that people don't mean to be offensive when they reference slavery this way. And you're right about that. Similarly, if I ask so…

Why is an analogy to slavery a bad thing? Such that even references to the concept need to banned

The Partition of India was a horrific tragedy and killed 2 million people. Does this mean we should do away with partition_sort ?

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

#903

Post by knorker »

bombcar wrote: post #351
trollied wrote: post #188

Earlier quoted context omitted.

Large planes tell pilots vocally to “retard” when landing…

Which can be amusing if you don't know what's going on, it suddenly sounds like the autopilot is pissed. https://www.youtube.com/watch?v=vmbzKsqKQoI And for those who would like to change this, any change to a pilot's routine has a chance of being the thing that pushes an incident over the edge into a crash. Would it be worth it?

I would predict that the same people who say we should defund the police (the ones that mean it literally), and say that "being on time", "work before play", and "be polite" is part of the white oppressive culture[1], and needs to be dismantled, would also argue that flying, and the methodology of the NTSB (and foreign equivalents) that has reduced airline accidents of the decades, is also part of the structural racism of white dominant culture.

I'm not saying this is a large set of people. But it looks like some of them have power at github.

I guess if you're doing software for the airline industry you cannot use github.

Well, honestly I wouldn't use github for anything if money depended on it, because of their record of "firing" customers who don't toe the line on their exact politics.

[1] I'm quoting training material that even passed HR review in corporate education DEI here. It's not hyperbole.

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

#904

Post by Elinvynia »

tempnow987 wrote: post #470
NtGuy25 wrote: post #354

Earlier quoted context omitted.

There's plenty of other companies that have the other side of the coin in values though that would admire people who stand up. It's very nice to be a good culture fit at a company. I have a Blue Lives Matter flag in the corner of my camera during interviews and it's gotten me some really good offers and I havn't been declined once. On the flip side, I know many that have been discriminated based on their looks or voc…

I actually would prefer that most political stuff stay out of the workplace. So minimize the crosses, the blue lives matter flags, the other stuff. In many of the more frugal businesses I've seen, folks just don't have the time to strike / protest etc for some of the stuff the FANG folks are into.

Not being able to strike is not the win you think it is, as striking is the most powerful action that labor has and has lead to the very things you enjoy today - weekends and a 40 hour work week.

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

#905

Post by suyjuris »

SamPatt wrote: post #787
words_and_signs wrote: post #743

Earlier quoted context omitted.

This is wrong. It's a good rule of thumb because passive voice can be a common indicator of weak writing, but it has plenty of good uses in strong writing. The best case to apply it is when the object of a sentence was previously a subject or other focus of prior sentences, and so is more familiar to the reader than the subject of the new sentence. For example: "The United States had maintained neutrality until 1941.…

"The United States was neutral until the Japanese attacked Pearl Harbor in 1941, leading them to enter the war." I'd say active is still better here.

I would disagree. This sentence is about the Unites States, not Japan, and the use of passive voice reflects that.

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

#906

Post by notahacker »

cuteboy19 wrote: post #901
chc wrote: post #890

Earlier quoted context omitted.

That seems like an odd example. The "specific intention" of the word "slave" in computing is unquestionably to draw an analogy to slavery, so if somebody objects to the word on the grounds that it references an actual atrocity, they're factually correct. I think what you're trying to say is that people don't mean to be offensive when they reference slavery this way. And you're right about that. Similarly, if I ask so…

Why is an analogy to slavery a bad thing? Such that even references to the concept need to banned The Partition of India was a horrific tragedy and killed 2 million people. Does this mean we should do away with partition_sort ?

Partition_sort isn't an analogy to the Partition of India (which isn't the traditionally understood meaning of the word 'partition').

In terms of analogies, we're much closer to if we'd decided to call a "delete all" function the "genocide" function. If computer programmers had widely used that terminology for the last three decades it might sound a bit less weird to us and the cohort of people shouting "that's such an inappropriate analogy" might be just a small group of liberals, but I'm not sure their argument about it being a poor word choice would be entirely baseless.

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

#907

Post by flippinburgers »

krastanov wrote: post #23
mgraczyk wrote: post #9

In the article they use the example of "Motherboard" and "Landlord" and seem to suggest that these are areas where the Google AI is making mistakes or being overly strict. As a Google employee expressing my own opinion and observations of company culture, I can say that these are 100% not mistakes. Many Google employees are just so out of touch with the real world that they believe it is the duty of Google Docs to ch…

I am incredulous at the "mother" to "birthing person" requirement. I have seen one or two people with such prescriptive views on language but the vast majority of people I know consider it ridiculous. And I am in an incredibly "woke" social bubble.

I think it is so extreme as to be just another calling card to "prove" your obedience to the cause. It is dystopian.

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

#908

Post by yanderekko »

ZeroGravitas wrote: post #893
rayiner wrote: post #746

Earlier quoted context omitted.

You can’t overlook the role that white progressives play in selectively amplifying minority voices. I can’t get on CNN to give the “Bangladeshi guy take” on rising crime by saying stuff most Bangladeshi people agree with. I have to provide a Bangladeshi perspective that is appealing to the white progressives that make the decisions at CNN. There’s a deep synergy here that exploits the fact that most Americans have li…

So, progressive groups support minorities, and then promote minorities with progressive opinions, and you don't like that because you're a minority with non-progressive views and feel underrepresented? I'm not sure I see the problem here? If there is a problem isn't it because minority groups that support regressive politics (say like the 'log cabin Republicans') don't get promoted by their fellow regressives as much…

The problem lies in the question of whether one can truly "support minorities" if you generally disregard their views and culture and only elevate their voices insofar as they agree with your politics. With conservatives do this, they're accused of tokenism and pandering.

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

#909

Post by indy »

noduerme wrote: post #865
epivosism wrote: post #809

Earlier quoted context omitted.

This guide is interesting in how it lists allegedly immoral language use, alongside with clear PR/legal/trademark directives which I don't give a fuck about, such as "don't use google as a verb". I'll google whatever I want, google PR, go suck an egg. Also quite annoying how they simply tell us what to do, without justifying things. A lot of these would be quite hard to truly argue - i.e. "blind people... can't see t…

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

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

#910

Post by selfmodruntime »

alanh wrote: post #755
yeetsfromhellL2 wrote: post #546

Earlier quoted context omitted.

Actually I've seen an English linter to help you maintain a passive voice for papers, remove waffle words and unnecessary fluff, etc. I'm not sure I can find the exact one I'm thinking of, but wasn't too bad overall, it spotted errors and made helpful suggestions. It was cool too, because it would read from stdin and integrated into vim pretty well with a few lines in my config.

> maintain a passive voice surely not? even a newspaper should be striving for an active voice, at least anytime the actor is known!

It would work for academic papers though
Post reply on HN