'Our findings reveal that AI systems emit between 130 and 1500 times less CO2e per page of text generated compared to human writers, while AI illustration systems emit between 310 and 2900 times less CO2e per image than their human counterparts.' SSRN

6mo ago I did a mini DC tour asking policy wonks why they were skeptical of AI, many said stuff like “ChatGPT has no common sense, if you ask for walking directions to the moon it’ll answer instead of saying it’s impossible.” Often they were thinking of much weaker/older AIs.

Just start seriously working on energy access already.

(Energy) poverty is very bad for humans and the environment: africanistperspective.com/p/energy-pover…

Super interesting!

'Why Small Nations Surprisingly Succeed' by Danny Quah

dannyquah.github.io/Storage/2024.0…

In *other* @cgdev news (ahem) from yesterday 🥁:

Rory Todd and I are out with a new paper and blog on methods for measuring lead exposure -- and why they're not quite up to snuff to meet most needs in LMICs 👇

cgdev.org/blog/why-world…

The US immigration system continues to have more weird quirks than I imagined (honestly, mostly a failure of imagination on my part)

Instant subscribe!

Upzoning doesn’t increase housing supply if the government attaches onerous affordable housing requirements.

This didn't happen. warhistoryonline.com/world-war-ii/m… (h/t Ruben C. Arslan)

“The mixing up of the words cryptogamist and cryptographer may have started as a joke among the staff at Bletchley Park.”

New article by me:

The rise in reported maternal mortality rates in the US is largely due to a change in measurement.
ourworldindata.org/rise-us-matern…

The change was adopted by different states at different times, resulting in what appeared to be a gradual rise in maternal mortality.

Nice

never gets outdated

Boston-area recipient of gene-edit pig kidney, Rick Slayman, is dead at 62, a month after leaving the hospital.

“We have no indication that [Slayman’s death] was the result of his recent transplant,” Mass. General Hospital said in a statement. 🧐

bostonglobe.com/2024/05/11/met…

Internal migration data in Brazil shows that common estimation procedures underestimate the importance of expected wages in migration choices. Workers face substantial and heterogeneous information frictions, from Porcher, Morales, and Thomas Fujiwara nber.org/papers/w32413

Does work from home work for women in India? an RCT that assigned jobs shows 56% of wives started working from home, while only 27% took office jobs. Higher wages did not affect this. Enforcing office check-ins reduced job take-up Suhani Jalota Lisa Ho shorturl.at/lJK16

+1! nytimes.com/2011/12/06/opi…

I liked this post by Lukas Gloor analysing prediction failures about Covid and AI.

It seems at least these two examples display an underestimate of people's reaction to danger, i.e. sleepwalk bias.

I think it's a general issue in the AI risk debate.

lesswrong.com/posts/dLwo67p7…

New paper in Ergo, with Teru Thomas: Suppose you don't care about the *total* amount of welfare in the world, but only about the *average*. Should you still care a lot about the size of the future population/long-term survival of humanity? Turns out, yes! journals.publishing.umich.edu/ergo/article/i…