Seema Jayachandran

@seema.bsky.social

Economics professor at Princeton. I study environmental conservation, gender equality, and (too many) other topics in developing countries. seemajayachandran.com

What appeals to me about Zohran is that he governs with a strong moral compass, even where I don't agree with the policies. The discourse is affordability vs. fight vs. gerontocracy. But I think the lesson for 2028 is the hunger for a moral compass.

This politician in Canada used ChatGPT to write his speech and then just directly read out loud the AI prompt without realising it, announcing to legislators “Here's a more natural, flowing version of that section that reads like a legislative speech rather than a series of short points." Glorious!

Do the cost savings from removing recycling and waste bins from offices really outweigh the productivity costs? The centralized bins and slots are small -- the setup is designed for you to stop working and walk from your office to the centralized bin for each piece of trash. Seems inane.

What countries have you only been to in transit? For me, the ones I remember are UAE Turkey Senegal Rwanda Malaysia Bangladesh Sudan The last two are probably the oddballs on that list.

I feel like there's a growing gap in perception between people who use LLMs to code or do code-adjacent things and people who only evaluate LLMs on their ability to write prose. This might be a stretched analogy but it reminds me of the different "worlds" of the polarized US media environment.

This was prompted by some policy folks describing some findings and me being skeptical. So I read the two working papers (by a prominent economist) and was aghast at the authors' over-interpretation. Not great that the policy folks were credulous, but I get that. I place the blame on the authors.

Seema Jayachandran@seema.bsky.social · last mo.

A tool I hope someone creates: Assessment of how much a scholar overclaims in their papers when interpreting empirical results. Seems well within the skill set of AI, e.g., Refine. Useful for non-experts assessing evidence (especially working papers but even peer reviewed). + a deterrence effect.

A tool I hope someone creates: Assessment of how much a scholar overclaims in their papers when interpreting empirical results. Seems well within the skill set of AI, e.g., Refine. Useful for non-experts assessing evidence (especially working papers but even peer reviewed). + a deterrence effect.

There’s been a lot of talk in this race about what makes a "real man." A man does what’s right when no one is watching. He upholds his commitments to his family and neighbors. He doesn’t lie, cheat, & steal his way through life. Real men serve others. Weak men serve themselves.