Understanding how given groups are represented in the broader culture is still interesting of course, but it is not necessarily the same as understanding ground-truth differences in how groups understand the world themselves. 9/10
4) How do we know? LLMs are most powerful in studying hard-to-survey cultural groups or those with little prior data, but this is also where it can be hardest to know if the LLM is faithfully representing the beliefs of a set of people versus stereotypes on the internet. 8/n
3) LLMs “know” cultural psychology. They may be learning from the cultural psychology literature itself and then repeating back to us things that psychologists have already discovered (in other words, the extant scientific literature could be influencing model output) 7/n
2) Whose writing??? Being written ABOUT is not the same as doing the writing. LLMs may overrepresent stereotypes, especially where outgroup writings are more voluminous than writings from inside the group. 6/n
1) The internet is not the world. What info makes it onto the internet is shaped by many forces, such as literacy, power, and the rewards that come with posting. Bias in, bias out. 5/n
HOWEVER. LLMs are biased. 4/n
LLMs can reproduce cultural differences: although WEIRD by default, they can be asked to respond ‘as if’ they were a person with certain properties (age, politics, etc). And their answers are remarkably accurate. 3/n
Cultural psychologists have used compressed cultural artifacts (like Google Book’s nGram) to track cultural change over time. For instance, US culture shifted in viewing happiness as “luck” (in the 1800s) to more modern views. LLMs could let us do the same on a massive scale. 2/n
New article at TiCS: What can ChatGPT and other large language models (LLMs) tell us about people? As a collection of compressed cultural artifacts, LLMs potentially allow for the study of culture at a massive scale. But caution is needed. 1/n 🧵
Two million in Bluesky - please post a celebratory bird
I'm going through old photos and found this. The last time there was Goose Discourse I mentioned this evil goose that would come after me when I was a kid at this local lake. This is not that goose, but that is the same lake, and I am certain this is one of that evil fucker's descendants
With grad admission deadlines coming up, I’m hoping to recruit a grad student (or 2!) this year! If you’re interested in guns, culture, or novels/reading, please reach out - I’d love to hear from you and chat about Madison.
(Please share!)
Bob’s 1986 AER, from his dissertation, compared estimates from an experimental and an econometric evaluation of the same job training program.
It showed limits of econometric approaches and is one of the most influential validation exercises in labor.
www.jstor.org/stable/1806062
Does what we read shape our worldviews? People who read more literary fiction growing up exhibit more complex worldviews (eg attributional complexity, lower essentialism) as adults - @nickbuttrick.bsky.social
#SESP #psychology #socialpsyc
“What is it about fiction, about poetry, about narrative that makes it so powerful? …that makes philosophers think of mass murder?” - @nickbuttrick.bsky.social on Plato
#SESP #psychology #socialpsyc
I'm assuming goose content is welcome here?
