Ruben Arslan @rubenarslan@schelling.pt

RT @jamesbrandecon
🚨🚨New working paper! 🚨🚨
@IsraeliAyelet, @NgweDonald and I have a new paper out which uses GPT-3 to estimate consumer preferences. We show that GPT-3's responses to conjoint-like questions are internally consistent, consistent with econ theory, and of reasonable magnitudes! 1/

RT @salonium
New Substack by me, on the great success of chickenpox vaccination in the United States, and what the rest of the world can learn from it.

(Plus, lots of other stuff you might not know about chickenpox and shingles.)
https://salonium.substack.com/p/13-the-success-of-chickenpox-vaccines

RT @tkaiser_science
Bildungsausschuss be like
#WissZeitVG #Ichbinhanna
H/t @maltoesermalte

RT @laura_j_botzet
Oh entschuldigung, so besser? https://twitter.com/CaroWagner_in/status/1637155289957801985

RT @literalbanana
did no one look at this and go "weird, that's way too big to be real"? (it's on a 100-point scale) https://people.duke.edu/~dandan/webfiles/PapersPI/Commercial%20Features.pdf

RT @farid_anvari
🚨#preprint🚨 Inspired by a thread I saw a couple of years ago (see here, https://twitter.com/rlucas11/status/1351573480584699904?s=20) I started a collaboration to look at how much people's repeated self-reports (of affect and personality) might be biased by response styles. 1/5

RT @JuliusStoll
@PHuenermund @starkwatzinger Der Mietendeckel: Jetzt auch für Academia!

RT @lakens
I vividly remember teaching p-curve analysis in 2014 to PhD students when a student in the back loudly swore. I asked 'what's wrong?'. She said 'I just p-curved the studies I have been trying to replicate for the first 1.5 years of my PhD, and now I see why I could not.'

RT @fchollet
Always keep in mind, though: GPT-3 and GPT-4 were trained on the public ARC tasks and their solutions. The tasks are distributed as JSON files part of a public GitHub repo, which is of course part of the training data.

This is exactly why the *test set* is fully private.

RT @axpuig
GPT-4 (just released by @OpenAI) exhibits greatly improved abstraction capabilities, similar to those required to master the ARC by @fchollet.
It can even solve this 2D task that demand some generalization skills. (1/7)

RT @JeffLadish
"can you write me a game in python where I control a pong paddle on the right side of the field and the left side of the field is Conway's game of life"

Gif of the resulting game after some additional instructions: https://imgur.io/rUc1x1N https://twitter.com/AndreTI/status/1635801920223989760

RT @alexeyguzey
@leopoldasch exactly like humans

RT @emollick
🤯🤯Well this is something else.

GPT-4 passes basically every exam. And doesn't just pass...
The Bar Exam: 90%
LSAT: 88%
GRE Quantitative: 80%, Verbal: 99%
Every AP, the SAT...