Sam Carter

Recently unearthed my favorite tweet in the history of twitter, posting here so I never lose it again (hi @mosenkis.bsky.social )

How to Fix Foreign Aid USAID's former Chief Economist reflects on DOGE

www.statecraft.pub/p/how-to-fix...

This looks like a phenomenal resource!

“the food marked for destruction could have met the nutritional needs of every child facing acute food insecurity in Gaza for a week”

USAID Shutdown Costs Top $6 Billion, Internal Estimate Shows The estimate, included in a seven-page memo prepared for Secretary of State Marco Rubio, raises questions about how much President Donald Trump will ultimately save taxpayers by closing USAID and othe...

No words for this hypocrisy. These people sacrificed millions of human lives, decades of progress against the world’s toughest challenges, and our country’s soft power—all while COSTING Americans $6B.
news.bgov.com/bloomberg-go...

1. Trump's tax cuts overwhelmingly help the rich
2. His spending cuts are targeted at the poor
3. Tariffs cost the poor a larger share of their income

Roll these three regressive policies together and you get the largest redistribution from poor to rich in American history.

The anti-Robin Hood.

Welcome! You are invited to join a webinar: AI for Good? Evaluating the Impact of AI in Development. After registering, you will receive a confirmation email about joining the webinar. Organisations looking to use AI for development are concerned about the potential for errors and biases when using AI in high-stakes situations. In this webinar, a collaboration with the Center for G...

📅 On June 12, join @hanshengchia.bsky.social, Samantha Carter, & Farhan Abrol for a @voxdev.bsky.social event exploring the potential applications of AI. W/ @cgdev.org, @j-pal.bsky.social & @agencyfund.bsky.social.

Register! ⬇️
cepr-org.zoom.us/webinar/regi...

(2) David issued a clear call to action for economists. Given their focus on understanding how to improve the economy, and given that AI will definitely affect the economy "we just have to roll up our sleeves and get to work" on figuring out what the future will look like.

(1) The last few minutes focus on how we should think about evaluations of AI, given that the technology changes so rapidly. David suggests that, "to remain relevant, researchers must focus on identifying underlying mechanisms that outlast any one model or interface."

Excellent podcast summarizing both a new RCT (tinyurl.com/findingtalen...) and the general discussion on "AI for the Global South." Two things I particularly liked:

The sweet taste of a new idea - MIT Schwarzman College of Computing Behavioral economist Sendhil Mullainathan has never forgotten the pleasure he felt the first time he tasted a delicious crisp, yet gooey Levain cookie. He compares the experience to when he encounters...

“Behavioral economist Sendhil Mullainathan has never forgotten the pleasure he felt the first time he tasted a delicious crisp, yet gooey Levain cookie" is an excellent lede

computing.mit.edu/news/the-swe...

A matchmaking service with a twist: Connecting big givers to programs cut by USAID A former USAID worker has a new mission. She's hoping to connect philanthropists with overseas programs that have lost — or are likely to lose — their U.S. funding.

USAID cuts have left major aid programs without funding. @cgdev.org is now hosting a team analyzing which ex-USAID programs are most cost-effective—and connecting them with donors willing to step in.

Learn more in @npr.org:
www.npr.org/sections/goa...

If it looks like the intervention you evaluated didn't work, what does that really mean? blogs.worldbank.org/en/impacteva... My summary of a taxonomy by Glewwe & Muralidharan #throwbackthursday

Quite the quote.

Great interview with @deankarlan.bsky.social! On generalizability (it matters but it's solvable), whether RCTs really are too slow (no, the eval method doesn't determine timelines, the outcomes of interest do), agency (exciting, need more research), etc.

If you find yourself confused by, tired of, and/or wanting more conversation about "evaluation" in AI, fear not! A new, practical framework from @theagencyfund.bsky.social is here. Hans @cgdev.org and I love it, and are helping to spread the word!

"Redirecting investments toward solutions that directly benefit society is essential to keeping AI development aligned with collective progress, rather than shareholder value."

Just came across this piece, and agree with it wholeheartedly! "Governments must realign incentives [from the current focus on automation] to encourage AI that serves social needs, such as enhancing education, improving health outcomes, and tackling climate challenges."

@justsand.bsky.social @charlesjkenny.bsky.social

What do we get wrong (i.e. misperceive) about the world? Why does it matter? How does fixing misperceptions improve society?

Research on @voxdev.bsky.social has answered all these questions and more, documenting a series of misperceptions that can be fixed 🧵

3 ways you can help the people hurt by Trump’s foreign aid cuts The president took away life-saving aid. You can give some of it back.

Worried about the people hurt by Trump’s cuts?

@sigalsamuel.bsky.social has curated some great resources for how to help:

Abhijit Banerjee and Esther Duflo: Foreign aid can be effective without the US In the wake of USAID cuts, wealthy nations and individuals must take the reins

“The world must now affirm that human lives matter, no matter where they are.” In a new op-ed, Abhijit Banerjee and Esther Duflo call for governments, multilaterals, and philanthropists to double down on investing in evidence-backed programs that improve people’s lives. www.ft.com/content/43e9...

Your tax dollars did amazing things.

Another op-ed by someone whose ACTUAL JOB was making US govt. more efficient, for everyone's benefit. "Our work was about trying to learn how to make government work better and more efficiently. I would have been very open to investigate the questions they had, but I didn’t have an opportunity to."

Importantly, nobody sees AI targeting tools as replacing conventional, ground-truth data like in-person surveys. But it's absolutely an important tool in efforts to ensure that aid reaches those who need it most.

The Togo example used data from satellite images and cell phone networks. @marshallburke.bsky.social and co-authors are also working on leveraging satellite data to predict poverty (4/n)
siepr.stanford.edu/news/harness...

That last piece is key: AI is not perfect (bias, exclusion), but it's often better than the other options that are actually on the table. Per Ariel BenYishay: "The baseline isn’t perfect data. It’s actually very crappy data." (3/n)

Using AI for targeting allowed Togo'"to leapfrog the conventional hurdles of using old and incomplete data" when providing COVID-19 relief, and reduced exclusion relative to available options (2/n) www.povertyactionlab.org/project/usin...

Can AI help beat poverty? Researchers test ways to aid the poorest people Measuring poverty is the first step to delivering support, but it has long been a costly, time-intensive and contentious endeavour.

How can AI be used for social good? A new piece in Nature highlights a few examples (1/n)
www.nature.com/articles/d41...

"We now have a million million tragedies that could have been avoided."