189 Photos and videos
Hello! I'm boosting a project I've been working on for a few months now---CalBallot---to get folks to try it out and let me know about their experience.
In a nutshell: CalBallot provides comprehensive coverage of state and local ballot measures spanning the last 30 years. (1/n)
1
2
281
still more of a nugget than a fully-fledged theory (may write on this eventually), but:
AI *feels* like a democratizing technology in a very broad sense while still having the potential for very negative/un-democratic outcomes in specific contexts.
defining those specific scenarios getting to workable solutions relatively quickly will be very important.
1
64
increasingly, I've become very interested in when things should happen but don't.
this is part of the motivation behind one of my dissertation chapters, which looks at why direct democracy (e.g., ballot measures), which puts questions to voters rather than have politicians decide, doesn't seem to produce a more informed public.
1
45
a lot of this probably has to do with incentives and information costs (see e.g., recent work in poli sci showing how it's mostly a lack of knowledge about candidates, and not administrative hurdles, that seem like the biggest barrier to participation)
but obviously a lot of potential for bias (prejudice?) againt tools as well. "I don't use AI (for voting, for work, for making art, etc.)"
lots to look into/learn!
19
sure as heck feels like it!
Jun 11
Not a cloud in the sky across virtually all of California today.
What a view from space.
59
1/ Happy to share that "Factorial Difference-in-Differences" (FDID), with Anqi Zhao and @pengding00, is out in JASA - ACS. doi.org/10.1080/01621459.202…
It has been a truly thrilling experience working with Anqi and Peng.
6
57
213
22,880
New blog post(s) coming soon-ish. Feel free to catch up on the existing ones if inclined!
igorgeyn.com/blog/index.html
30
very cool article that nails both the "data" and the "journalism"
chicagomaroon.github.io/data…
55
Igor Geyn retweeted
May 20
People who care about AI: Free Systems needs your help! What would you like to know about the capabilities, dangers, and other characteristics of cutting-edge AI models?
Inspired by a suggestion from @Miles_Brundage we are working to design a simple, clean model card visualizer. The goal is to give people a quick, easy way to see what's different across models and what's new in the latest models.
But it turns out, it's pretty hard to boil down model cards to a single actual "card." New model cards can be many pages long. Most of the data provided in model cards is not directly comparable to other cards (there's almost no overlap in evals, we've found).
So we want to know: what are you looking for in a model card? What are the pieces of information that would be most valuable to you? And how would we standardize these cards across labs and models?
If you have a moment, please fill out our survey here:
forms.gle/9k6E5e11Z4WNUim88
We're hoping to ship our visualizer two weeks from today, and would really appreciate your input.
3
5
20
2,638
underrated (IMO) thing:
the status quo of things like transferring phone carriers (outside of the typical/most common case), getting a car registered from out of state, etc. is a *huge* disaster and time suck,
and the government should do something to help people avoid spending countless hours on economically useless activities.
1
54
(putting on the PhD hat for a bit, or--more accurately--failing to take it off when the situation may not call for it)
It does seem like the incentives/intuition about where efficiency is created is more interesting than meets the eye. The DMV has gotten *a lot* better over the 2-5 years (IME). mostly through technology.
almost every commercial app/tool that I've been using for 5-10 years has gotten worse.
98
Sharing a few analyses I've been working on as California's (and other states') wildfire seasons roll on. Already some big blazes this year---hoping for a safe season for all.
linkedin.com/feed/update/urn…
29