J.D. Porter

New editorial from Emily Hammer & the @pricelab.bsky.social Critical AI Working Group at Penn. "The Penn administration now budgets huge sums for licensing AI products. Our leaders should devote equal resources toward new physical spaces and staff that support innovations in AI-free instruction."

JCO is making a (dubious) sociological claim here rather than a technical one (“some kinds of readers don’t care if what they read is by people”), but I do think it's interesting that, on the level of the actual text, LLMs are probably a lot better at simulating literary fiction than romance novels.

I think JCO is saying that even if LLMs could make both kinds of books, romance readers would be more likely to accept them. She sees romance readers as indiscriminate consumers of the novel-as-commodity. This seems to me totally belied by the incredible complexity of romance as a genre system.

JCO is making a (dubious) sociological claim here rather than a technical one (“some kinds of readers don’t care if what they read is by people”), but I do think it's interesting that, on the level of the actual text, LLMs are probably a lot better at simulating literary fiction than romance novels.

Consider these two books: 1) A rambling meditation on modern labor written in an uncanny corporate prose style 2) A tightly-plotted love story with compelling characters Which seems easier for an LLM to make? Which one is “literary” and which a romance?

I saw this painting today (“Seine”, by Ellsworth Kelly), and I’m increasingly convinced that modernism holds an answer for us about how AI art will be received if it ever becomes pretty indistinguishable from human art.

This is an offshoot of the famous “my kid could do this” response to abstract art. Most kids couldn’t, sure, but the more accurate response was always: “do what?” The answer to that cannot be automated.

I saw this painting today (“Seine”, by Ellsworth Kelly), and I’m increasingly convinced that modernism holds an answer for us about how AI art will be received if it ever becomes pretty indistinguishable from human art.

This is part of what modernism worked through! The work of art derives its meaning not from form alone but from an entire context. Duchamp must pick the particular toilet and put it in an art show. People must interpret it. Otherwise it’s just a toilet.

I saw this painting today (“Seine”, by Ellsworth Kelly), and I’m increasingly convinced that modernism holds an answer for us about how AI art will be received if it ever becomes pretty indistinguishable from human art.

But this has been true in modernism for over a century. Duchamp’s “Fountain” was literally mass produced. Modernist poetry like this isn’t random (I love this poem) but very simple algorithms could easily make a “poem” that an ordinary reader would struggle to differentiate from it.

I saw this painting today (“Seine”, by Ellsworth Kelly), and I’m increasingly convinced that modernism holds an answer for us about how AI art will be received if it ever becomes pretty indistinguishable from human art.

An AI “actor” or “performer” is really just a cartoon. They don’t have any more agency or interiority or personhood than Elmer Fudd.

Trump has tried to replicate each other President’s signature failures, from the hits (internment camps, corruption, losing a war in Asia) to the deep cuts (tariffs, destroying the WH, picking a contemptible racist VP), but I’m still kind of surprised to see him going for the William Henry Harrison.

This is obvious horseshit, but I think it's useful to put numbers to this stuff sometimes. The Warren wealth tax would raise $16.5b from Bezos alone in the first year, which amounts to $4k for every teacher in the US, roughly a 6% pay raise on average. Bet that *would* help that teacher in Queens!

Here's the math: Warren tax = 0 on first $50m + 2% on next $950m + 6% on the rest Bezos is so rich (net worth ~$275b) that only the last number matters, and it comes about to about $16.4b. There are 3.8m teachers in the US, making on average $74k/yr. 16.4b / 3.8m = ~$4.3k 4.3k / 74k = 5.8%

How bad a foul baiter is Shai Gilgeous-Alexander? See if you can guess which two dots are his on this graph of MVP seasons. Dots are further right if players shot the ball a lot. Higher dots mean players shot more free throws. Answer lower in the thread.

One thing I hadn’t realized until I made this is that Jokić takes relatively few FT, especially for a big man. My theory is that it’s because he’s usually just standing there when he shoots. Walton, another floor-general big, is an interesting comp.

How bad a foul baiter is Shai Gilgeous-Alexander? See if you can guess which two dots are his on this graph of MVP seasons. Dots are further right if players shot the ball a lot. Higher dots mean players shot more free throws. Answer lower in the thread.

Bam Adebayo now holds the title belt for highest scoring game among active players. He's just the 34th player with this achievement, and the 21st since Elgin Baylor (the first good offensive player).

When Robinson retired 9 years after that, Karl was still around, and got the title back (now tied with Shaq) for that same 1990 game. Finally TMac got rid of him for good by scoring 62 in 2004.

Bam Adebayo now holds the title belt for highest scoring game among active players. He's just the 34th player with this achievement, and the 21st since Elgin Baylor (the first good offensive player).

The first title holder was Ed Sadowski, who scored 18 points for the Toronto Huskies in the first BAA game. Some pretty amazing names in that first month: Max Zaslofsky, Mel Riebe, Dino Martin, Leo Mogus. Joe Fulks finally stabilized the record via preposterous ball-hogging (63 points on 56 shots).