Zum Inhalt der Seite gehen


Microsoft Research is at it again:

Advait Sarkar, a Microsoft Research employee got a paper published at the CHI Conference on Human Factors in computing systems:

"AI Could Have Written This: Birth of a Classist Slur in Knowledge Work"

dl.acm.org/doi/10.1145/3706599…

Now I am all for calling out structures and language of oppression and discrimination but this is really something special. "AI shaming arises from class anxiety in middle class knowledge workers" ... yeah no. It's about shaming people (usually from the middle or upper class) who don't want to put in actual work but get credit for having done it.

Like what is the argument: Lower class people can only compete while using AI and therefore should not be shamed? What kind of a view on lower class people does Mr. Microsoft Research communicate here?

teilten dies erneut

Als Antwort auf tante

Do I wanna read this paper to critique it better or is this just an absolute waste of time?
Dieser Beitrag wurde bearbeitet. (3 Wochen her)
Als Antwort auf climbertobby

@climbertobby I mean it's rare that one gets dumber reading papers. Question is what you expect you'll learn
Als Antwort auf tante

@climbertobby I wanted to read it but do not want to raise their stats.. Do they have data (even qualitative/corpus)? What are their background theories ("limit mobility into knowledge work" sound more management than sociology)?
Als Antwort auf Olivier Leroy

@defuneste "boundary work" mentioned in the abstract sounds more like sociology. But from the abstract alone, I can't even see if they use the term correctly
Als Antwort auf climbertobby

@climbertobby @defuneste youtube.com/watch?v=Rir-HoYJNY… here's a ten minute video of him presenting his paper
Als Antwort auf tante

@climbertobby will try to watch it when I am not in coworking/open space ;) thanks!
Als Antwort auf Olivier Leroy

@defuneste @climbertobby
The paper is on the author's website.

Reading it gives you a strange feeling of being affected by something like reverse gaslighting.

I think this is due to the fact that it uses several conceptual double binds (like making a moral argument while at the same time decrying codes of ethics of AI use as "moralisation") and completely ignores the economy of AI proliferation.

In the paper, the technology is simply just there. Enormous investment of capital and corporate landgrab play no role whatsoever while discussing "classism".

The only time they get vaguely hinted at is when towards the end possible disenfranchisment is dismissed with the claim that there will supposedly soon be a perfect market where producers have little power anyway.

Als Antwort auf Vogelgesang

@foyglgezang @climbertobby

In french I would say "kamoulox" but woah, this was published????

Let me ignore any kind of art theories, let me ignore any kind of materialism (But I will still say "Marx") because "these arguments cloak a deeper underlying motivation for
shaming, of which may the shamer may not even be conscious."

We are so lucky to have the author and the revealation! I will need to calm done after the use of Foucault. ....

Als Antwort auf Olivier Leroy

@defuneste @climbertobby Well "published"... There are hundreds of papers / extended abstracts in the proceedings of this conference, and Microsoft was one of the sponsors.

My favorite part is when the author recalls having read a paper about "delve" being a marker for AI use and gives the associated "culture of shame" the name of a biblical god that demands child sacrifice.

Dieser Beitrag wurde bearbeitet. (3 Wochen her)
Als Antwort auf Vogelgesang

@foyglgezang @defuneste @climbertobby He did present it in a session (there's a recording) so it wasn't just "sponsor can add some garbage to the proceedings", it was at least treaded as if it were a real contribution
Als Antwort auf tante

@foyglgezang @defuneste @climbertobby CHI is also considered by many to be the most prestigious venue for publications in the field. Many researchers submit their best work there. There are 1000s of publications because it's a big dysfunctional Moloch with a broken review process that is unhealthy for the HCI community and should be divided more meaningfully in smaller parts (or, even better, replaced by journals).
Als Antwort auf tante

The paper was in the alt.chi track which is meant for contributions that are "controversial" and would likely not get accepted in the regular paper review process. chi2025.acm.org/for-authors/al…

Submission requires you to also hand in a presentation video, I think that's the one on YouTube. Means it was maybe not recorded in a presentation at the conference proper, but rather beforehand by the author? [Edited:] Might also have the audio track from a live presentation there, I am not sure - I think you are right, it was likely presented.

Anyway there was a selection and review process. Since the submissions are not anonymous I think it is reasonable to expect a certain bias towards accepting at least some submissions by authors affiliated with a major sponsor.

Dieser Beitrag wurde bearbeitet. (3 Wochen her)
Als Antwort auf tante

@climbertobby @defuneste I watched the video in it's entirety. I would like to believe this is a practical joke and not just straight up, highest grade bullshit.
Als Antwort auf tante

@defuneste The more I listen to this, the more this seems like AI-hype bullshit with the false promise, that using AI can provide upward class mobility.

Also the term "boundary work" is used with plenty of liberty if not just wrong.

Also I bet that guy would interpret Marx as saying Capitalists act immorally.

AI shaming is sexism against men, sure mydude, sure 🤡

Als Antwort auf tante

(This is the short version as his presentation at the conference. The argument is ...
not exactly rigorous I'd say
youtube.com/watch?v=Rir-HoYJNY… )
Als Antwort auf tante

I haven't read the paper but I've heard the classism/AI hate argument before and it usually goes like this:

Rich people are more likely to get their papers accepted, not because the content is better, but because they went to good schools so the way they express themselves is different, and editors and reviewers pick up on that.

AI is good at mimicking that writing style, so
poor people can now sneak into the club by running their paper through AI to poshify it. Hence the sneering.

Als Antwort auf tante

an AI could’ve written this paper (it probably did)
Als Antwort auf tante

Oh, is this like when Musk got rid of the old "classist" Twitter verification system (with "lords and peasants") and replaced it with checkmarks ("for everyone"!) that cost $$$ but are functionally useless (because no verification happens anymore)?
Als Antwort auf tante

there is an actual argument for this, or at least there was (AI writing characteristics may have changed since so I don't know how much it still applies).

Basically since LLM AI was heavily trained by underpaid workers in India and Nigeria among other countries, some of the "tells" of AI writing were actually Indianisms and Nigerianisms. Which meant that when Indian and Nigerian people wrote naturally they were more likely to be falsely accused of getting AI to write for them.

Als Antwort auf tante

EPS radar plants flags early
90% hit rate → Quarterly profit exceeds 60%
💣 MSFT/GOOG targets tonight
🗺️ 100 maps available
🔗chat.whatsapp.com/F618WpgwKN7L… 👈
Als Antwort auf tante

I don't know what you are saying here.

'shaming people (usually from the middle or upper class) who don't want to put in actual work but get credit for having done it.'

So you are critical of people who use AI and take credit for it? Or critical of managers who use employee work and take credit for it?

'Lower class people can only compete while using AI and therefore should not be shamed?'

Are you suggesting that we *should* shame lower class people who use AI?

🤔

Als Antwort auf Phosphenes

@Phosphenes We should shame everyone for using "AI", with preference for middle- and upper class people. Using "AI" is shameful.
Als Antwort auf tante

This had me in hysterics (anad it's perfectly time for something I'm writing), so thank you! How is this different from earlier forms of plagiarism, when people would take old essays and duff them up a bit?
Als Antwort auf tante

The other way to look at this is: Microsoft is so worried about public perceptions of AI that they're paying researchers to construct arguments for why it's unethical to speak ill of their product.
Als Antwort auf L⁂Rhodes

@lrhodes Not just a researcher but one associated with the University of Cambridge.
I don't think it's a planned thing by MS though. Just some boot licking.
Als Antwort auf tante

That paper is quite something. What strikes me as particularly devious is how he uses "class" in the liberal sense as a synonym for milieu or social setting and not in the marxist sense. By doing that he actually manages to de-politicise "AI shaming" and reframes it as an issue of envy within the exploited class.
Als Antwort auf tante

This is pure, unadulterated horse shit. I’ve heard this same foolishness many times before from AI boosters.

“Criticizing someone for using AI is just a gatekeeping; you’re trying to protect your ‘privileged status’ as a developer/artist/etc.”

Als Antwort auf tante

I have come to the point of automatically deciding that anybody shilling LLMbeciles is either a complete and absolute moron, or an ethics-free sociopath.

There's a lot of both of those out there it turns out.

Als Antwort auf tante

Yes its an Alt-CHI track paper which are meant to be provocative, and no, Sponsors do not get preferential treatment in acceptances

David England, CHI participant since 1987