Mamdani agrees to ban NYC horse carriages, sealing industry’s fate
https://www.crainsnewyork.com/politics-policy/cny-mamdani-menin-council-deal-horse-drawn-carriage-ban-20261013/
begon, nigger.
so carney (allegedly) posts on linkedin? told you it was comp'd comp'd comp'd
when these good folks get radicalized, it's not going to be big fun
enough with the slut vibe already.
DANCE!!
drawing stuff with planes
Giving an LLM someone's entire attitude profile instead of just their age and gender barely helps it guess how they react to a post.
About two percentage points better… on average. And the average hides the stranger result: a plain word-counting classifier from the 1990s beat every model tested.
A new Nature article surveyed 1,511 people in Serbia, had them react to 56 social media posts, then built personas at three levels of detail and fed those to 27 frontier LLMs, generating around 6.85 million simulated reactions to compare against the real ones.
Full personas got 71.5% of reactions right vs 69.6% for demographics-only prompts. Once you correct for chance agreement (0 is coin-flip, 1 is perfect), GPT-5.2 came out at 0.30 while a bag-of-words logistic regression trained on the same data hit 0.36, and its probabilities were far better matched to reality (0.20 vs 0.33 on a lower-is-better calibration score). The models also guessed positive reactions correctly 68% of the time and negative ones only 55%.
Looks like most of the "digital twin" panels being sold right now are effectively running on the demographic line of the prompt and treating the rest as decoration. Which model you pick apparently matters more than how much you know about the person… and a logistic regression from the 90s might beat both.
https://lnkd.in/p/eXWJWsKg
anons, didja know you have "digital twins"?
reeling that one in, tunafish. been looking for it rong time.