Lol I would love to think so, though I suspect some bad frames are a symptom rather than a cause.

But if it helps regular folks stay focused, ignore bullshit, and communicate clearly, that's important and useful and I'm glad of it

i mean this sincerely: i respect your commitment to intellectual rigor

you're welcome / i'm sorry

i suggest using "predictive models" which leads nicely to "what is the model predicting"

when you think you're all alone,
google is listening

when there's nobody home,
google is listening

sharing secrets in the dark
with your lover, or your friends,
google is listening, google is listening, google is listening

in the middle of the night,
when you've got questions with no answers
google is listening, google is listening, google is listening

i think this holds true for a lot of tech stuff. harms are always presented as risks, because if you admit they're already happening it really changes the vibe from "hypothetical discussion of possible futures, maybe good ones!" to "how will you address these real harms now"

when you need to talk to someone you hold dear,
google is listening

the first xkcd

100% like just put it in the real context: this is a bad faith attack, there are more of them all the time, and treating it as a real / valid concern is why we keep getting more of them

you can never curl up into a defensive position that bad faith can't attack, bc they'll just make things up

it's wild bc frame negation kirby would be

frame: this prof gave a discriminatory grade
negation: actually the essay sucks and deserved a zero
kirby: this is a bad faith attack on a trans grad student

but a lot of people aren't even negating, they're saying "well maybe they were too harsh"

right the ai flavor is not changing anything important, it's just there to be like "but now we have a computer that says so"

good for a ribbon cutting perhaps

"not our models, our models are right, you're wrong" -ai boosters

because if a model is a tool for helping us understand something, then our ability to understand why it predicts something is central to whether it's a good model

Like ai fails horribly as a tool for actually doing useful work, and it succeeds horribly as a tool for justifying things.

But we can make it fail at that too!

Oh for sure. And NPR is just a useful example of a kind of credulous reporting that you see everywhere

30 minutes away is just close enough that people there probably already have rude nicknames for his town and the people in it

Like does their model account for the way "I'm running to represent my community where I live" reads as authentic

whereas "I'm running to represent a district I don't live in bc the computer said i would win" reads like a soulless hack controlled by consultants