Since Aristotle, literary theory has studied texts and reception. We propose a new object of literary study: the latent spaces of an AI model trained on a single text. We call the method latent reading and apply it to the first case study: FinneganLM, trained on Joyce’s Finnegans Wake.

Re: Last retweet, I will say I was impressed by the reviews from @aiide.bsky.social. And not even because they were positive (I think our paper probably won't get in). But they clearly read the paper carefully, were familiar with both our and other related prior work, and had very useful feedback.

VLADIMIR: Have they responded to our rebuttal? ESTRAGON: I haven't checked. VLADIMIR: You must. ESTRAGON: Why? VLADIMIR: To give us hope. ESTRAGON: Ok. We have a response. VLADIMIR: And? ESTRAGON: "I keep my score." VLADIMIR: (Stunned) He keeps his score? ESTRAGON: He keeps his score.

Melanie Mitchell@melaniemitchell.bsky.social · 2d ago

Edgy update of Waiting for Godot: Vladimir and Estragon are NeurIPS authors waiting for any reviewer to respond to their rebuttals.

Assistant or Associate Teaching Track professor opening at Carnegie Mellon's Entertainment Technology Center (ETC)

They've outdone themselves with this one. Appears to be from a game in the Monkey GO Happy series?

Mark J. Nelson@mm-jj-nn.bsky.social · last yr.

My apartment management continues to make befuddling use of clip art. Definitely not generative AI but somehow the mistakes are worse.

Noticing a pattern across departments that faculty are generally a lot more excited about AI than students, esp undergrad students.

Looks pretty interesting!

Nature@nature.com · 2w ago

Nature research paper: People use fast and flat simulation to reason about new games go.nature.com/4vxhiGA

I'm honestly happy to see some film-related Discourse on this site. I don't watch films myself, of course, because it's 2026. But I do think it's important that all the weird 20th-century retro media forms have communities striving to keep them alive.

This is kind of old news, but I pulled up 'macmon' on my Macbook, and still find it amazing that I can be running a full-featured laptop with a bunch of stuff open, and the total power draw is... 3 watts. Like 1/20 of what one single medium-brightness lightbulb used to use.

the litter robot's return to service draws substantial customer interest

Not a strongly held opinion, but I'm a little skeptical of the recent LLM math results not really being compared against baseline search methods with similar compute budgets. Some of them are using pretty huge compute budgets!

I keep an approximate time log so I have an idea what I'm spending work time on. My one-semester gig as interim department chair shows up very clearly!

Kind of reassuring to read that one of the biggest current problems in ML for drug discovery isn't any kind of exotic new AI/ML problem but just, still, the difficulty of preventing data leakage from the test set.

Benchmarking the Benchmarks

science.org

I'm not normally a morning person, but I'm catsitting some cats who are used to having breakfast before their small human leaves for elementary school, so I guess I wake up around 8am now.

What I mumble to myself whenever making a decision.

Completed my @aiide.bsky.social reviews, only a few hours late. 😅 (They were due yesterday "AOE time", which I think was 8am EST today.)

Recently discovered that the founder of our psychology department got fired after only a year due to a mail-fraud indictment. Later went on to a successful career in Hollywood, where he created Wonder Woman.

can't believe Big Mosquito is astroturfing my neighborhood

I see people are worried again about the increasing volume of papers in AI/ML. Just wanted to re-up my revolutionary O(1) solution*: browse the literature in time independent of paper count! * Assuming an idealized UI model where we can trigger timers & refresh content arbitrarily quickly

Mark J. Nelson@mm-jj-nn.bsky.social · 7mo ago

I uh, made this. It was supposed to be a joke / concept-art thing that scrolls through the torrent of new AI/ML arXiv uploads too fast to read. But I think I iterated too much and made it almost usable.

spotted in the library stacks while looking for something else

On the off chance you want to hear me talk about intersections between games and the history of AI, I'm giving at talk at the Arlington Public Library on August 4. Though it's intended more for a general public audience, and likely won't have too much new for people who follow me here.

AI and Games: A Recent History

Discover how AI changed gaming, and how gaming changed AI Find out how artificial intelligence has been considered in light of video games--and vice-versa. Go beyond simulated...

arlingtonva.libcal.com

Siri, give me the name of an Italian mathematician

The second instalment of our series "Haskellers from the trenches" is about Enterprise #Haskell at H-E-B, the Texan grocery chain. Blog post: blog.haskell.org/enterprise-h... Discourse: discourse.haskell.org/t/enterprise...

Enterprise Haskell at H-E-B | The Haskell Programming Language's blog

How Haskell earned its place in H-E-B's supply chain.

blog.haskell.org

smoke mostly cleared up

A paper casting doubt on the foundations of a bunch of nice convergence bounds in multi-agent RL (found via @quokkka.bsky.social). The technical material here is a bit out of my depth, but relevant to some stuff I do in game AI. Attempt to briefly summarize based on an initial read below. 1/n

Paradoxes of Game Theoretic Equilibria and Price of Anarchy

For decades, static solution concepts (Nash, Correlated, and Coarse Correlated Equilibria) and the Price of Anarchy (PoA) have formed the bedrock of algorithmic game theory, with no-regret learning pr...

arxiv.org

hot day train ride

Woke up in a sweat at 4:30am, with one thought.

Mark J. Nelson@mm-jj-nn.bsky.social · 5mo ago

un-cancel the Hilbert program!!

Consider sponsoring the AAAI AIIDE conference in Belo Horizonte, Brazil! Your support will help motivate cutting-edge advancements in Game AI and creative technologies while giving you access to the best global talent in the field. Get involved: sites.google.com/view/aiide20...

AIIDE 2026 - Sponsors

Call for Sponsors We are excited to announce the Call for Sponsors for the The 22nd AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE 2026), taking place in Belo ...

sites.google.com

This criticism of early 2020s Netflix originals is fascinatingly similar to the complaints people have about genAI content. But all made by teams of paid human creatives following a formula.

Sad interactive narrative research news: learned via @hkoenitz.bsky.social on LinkedIn that longtime pillar of the community Frank Nack died last week.

This is the kind of post you never want to write. My close friend, collaborator, mentor, fellow ARDIN founder and fellow ARDIN executive board member Frank Nack passed in his sleep last Friday. He… |...

This is the kind of post you never want to write. My close friend, collaborator, mentor, fellow ARDIN founder and fellow ARDIN executive board member Frank Nack passed in his sleep last Friday. He ha...

linkedin.com