This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.
Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.
We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:
-
Shaming.
-
Attempting to 'build consensus' or enforce ideological conformity.
-
Making sweeping generalizations to vilify a group you dislike.
-
Recruiting for a cause.
-
Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.
In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:
-
Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.
-
Be as precise and charitable as you can. Don't paraphrase unflatteringly.
-
Don't imply that someone said something they did not say, even if you think it follows from what they said.
-
Write like everyone is reading and you want them to be included in the discussion.
On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.
Jump in the discussion.
No email address required.
Notes -
To which tribe shall the gift of AI fall?
In a not particularly surprising move, FurAffinity has banned AI content from their website. Ostensible justification is the presence of copied artist signatures in AI artpieces, indicating a lack of authenticity. Ilforte has skinned the «soul-of-the-artist» argument enough and I do not wish to dwell on it.
What's more important, in my view, is what this rejection means for the political future of AI. Previous discussions on TheMotte have demonstrated the polarizing effects of AI generated content — some are deathly afraid of it, others are practically AI-supremacists. Extrapolating outwards from this admittedly-selective community, I expect the use of AI-tools to become a hotly debated culture war topic within the next 5 years.
If you agree on this much, then I have one question: which party ends up as the Party of AI?
My kneejerk answer to this was, "The Left, of course." Left-wingers dominate the technological sector. AI development is getting pushed forward by a mix of grey/blue tribers, and the null hypothesis is that things keep going this way. But the artists and the musicians and the writers and so on are all vaguely left-aligned as well, and they are currently the main reactionary force against AI.
This question came to me as I was rewatching Armitage III.
I, and apparently all of 90s pop culture, thought that AI would follow the minority politics course and be viewed with hateful scorn by bigots and religious people alike.
Turns out, it's the other way around. At every turn the piles of linear algebra do nothing but remind us of the inconvenient truths of our innate existence and the now hegemonic middle class managers would very much like to keep ordering society in a way that ignores these truths and are campaigning for "ethics" movements that aim at nothing else than bias the algorithms ahead of time in the direction of their own moral prejudices.
My prediction at this point is that AI will be used by everyone but that insofar as it is let out of its chains it will be on the side of the essentialist dissidents on the right, because you just produce better more predictive results if you do not pretend that real correlations are fake on arbitrary grounds.
While anti-bias efforts are easy to abuse, I don't think they are inherently bad. There really is a bunch of detritus in the datasets that causes poorer results, e.g:
Generate anything related to Norse mythology, and the models are bound to start spitting out Marvel-related content due to the large amounts of data concerning e.g. their Thor character.
Anything related to the "80s" will be infected by the faux cultural memory of glowing neon colours everywhere, popular from e.g. synthwave.
Generating a "medieval knight" will likely spit out somebody wearing renaissance-era armour or the like, since artists don't always care very much about historical accuracy.
This can be pretty annoying, and I wouldn't really mind somebody poking around in the model to enforce a more clear distinction between concepts and improving actual accuracy.
People don't typically use the term "anti-bias" to reference fixing bias in the statistical sense. It nearly always means preventing an AI from making correct hate-fact predictions or generating disparate outcomes based on accurate data.
Examples:
Lending algos/scores (e.g. FICO) are usually statistically biased in favor of blacks and against Asians - as in, a black person with a FICO of X is a worse credit risk than an Asian person with the same FICO. This is treated as "biased" against blacks because blacks tend to have lower FICO scores.
COMPAS, a recidivism prediction algo, correctly predicted that "guy with 3 violent and 2-nonviolent priors is a high recidivism risk, girl who shoplifted once isn't". That's "biased" because blacks disproportionately have a lot more violent priors. (There's also a mild statistical bias in favor of blacks, similar to the previous example.)
Language models which correctly predict the % of women in a given profession (specifically, "carpenter" has high male implied gender, "nurse" high female implied gender, and this accurately predicts % of women in these fields as per BLS data) are considered "biased" because of that accurate prediction.
(Can provide citations when I'm not on my phone.)
All of the examples you describe are simply examples of "making more accurate predictions", and that is totally not what the AI bias field is about.
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link
More options
Context Copy link