Kevin Kloiber

@BeatrizGietner In the other game I posted, I restricted it to abstracts from 1990 to 2020, and I should have probably set an upper bound here as well. Good point!

@jt_kerwin @cblatts Thanks for playing! I went through some of them and thought I spotted most of the errors, but I guess I have to do some more thorough cleaning.

@EconKevin @cblatts Some of these are obviously real because of PDF parsing errors. Would not have done so well without them in the sample.

Handed in my thesis two weeks ago and finally have time for fun again. So I built two small games about AI vs. real economics writing. Can you tell a real top-5 journal abstract from an AI-generated fake? Three lives, no hints, leaderboard. kevin-kloiber.com/ai-or-real.html #EconTwitter

@CRo4st That's actually what a lot of economics research is about! Estimating causal effects, so "X increased Y by Z percent," is pretty standard in the field, even if not every abstract spells it out that way. Good point on the %, if the fakes are too predictable, something to tweak! :)

@EconKevin Very interesting. From somewhere, AI gets the idea that economics abstracts say "a made b increase by c%", but it absolutely can't be from real abstracts. #1 now

@JRAngEcon Interesting observation! I randomly pulled 500 real abstracts from top 5 so it was not designed in that way. I cut the abstracts after roughly 80 words, in order to not bore players, so they might have numbers in later parts.

@EconKevin Tried the first one and I noticed that abstracts with no numbers were almost all the time real and those with numbers were AI-generated. Was it designed that way?

@econ_lessmann @DinaPomeranz What I can tell is that it’s a) not my score and b) that’s it’s genuine as in not injected. But one could still Google the abstracts..

@DinaPomeranz My score: 3. Who the fuck makes 42? That must have been the same AI that generated the game.

I got 3/3 wrong...

@ZainHameedo @captgouda24 Odd but true. So version 2 will include bad flowing AI abstracts. ;)

@EconKevin @captgouda24 lol a trick might be that any language that doesn't flow right had to be human lmao

Both are still work in progress so happy to hear any feedback, bug reports, or ideas. And yes, I'm keeping an eye on the leaderboard.

Also built a second one, inspired by a recent NYT piece: two passages from the same section of an economics paper. One is human, one is AI. Pick the one that reads better and see how other readers voted. kevin-kloiber.com/choose-your-si… #AcademicTwitter

Very excited to share our new working paper on arxiv, co-authored with @CausalHuber and @lukaslaffers! Check it out! 🚀📚 #EconTwitter #CausalInference

Delighted to attend the European #CausalInference Meeting in #Copenhagen @TheEuroCIM, where I will present work (joint with Kevin Kloiber and @lukaslaffers) on testing identification in dynamic treatment and mediation models today. eurocim.org

Fast genau 6 Monate später nun die Gewissheit - auch in naher Zukunft werden wir nicht mit deutschen Gesundheitsdaten (ökonomische) Forschung betreiben können. Quo vadis, @bfarm_de @Karl_Lauterbach?

We closed the workshop with more great presentations today!!! Thanks to all participants for coming to @cepa_unipotsdam and contributing to this event🙏🏻. Feeling happy and exhausted😉. Safe travels home and have a nice weekend!

After 5 days, we wrapped up the CINCH-dggoe academy in #healtheconomics To all participants: THANK YOU 🙏It's been a terrific experience and great fun 🥳 For #phd students who missed out, follow us @CINCHessen -- we will return next year! 💪 #econtwitter @dggoe @unidue

Day 2 of the #CINCH @dggoe Academy in #healtheconomics Capping off a wonderful day of lectures and presentations with a great group of aspiring young scholars at the UNESCO world heritage site @ZecheZollverein Some impressions ⭐

Das neue FDZ Gesundheit im Aufbau. Ein Drama in 3(+) Akten. Traurige Wahrheit über Datenverfügbarkeit in Deutschland.

@Stata Economics

Cozy up with a new #Stata beanie and fleece pullover! To enter: • Comment below what field you’re currently in • Click the ❤️ below • Be sure you’re following us We’ll announce the winner at 10:00 a.m. CT on 22 March.

@Stata 2014!

New year, new merch! We’re giving away some #Stata goodies to celebrate the new year and semester. To enter: • Comment below the year you started using Stata • Click the ❤️ below • Be sure you’re following us. Entries close at 10:00 a.m. CT on 22 February

Looking forward to the workshop with many interesting talks and people!

Today and tomorrow: @ifo_Institut Center for Macroeconomics and Surveys Research Days w @bornecon @BachmannRudi @almutballeer @_peterzorn @EconKevin @Dominik__Sachs @ManuelMenkhoff @julia_baarck @KlausWohlrabe @MathiasDolls @GerWo2 @SimonFKrause