Chris

I feel like I'm on the 0.5-speed track and it still feels way fast.

Lessons from a Chimp: AI "Scheming" and the Quest for Ape Language We examine recent research that asks whether current AI systems may be developing a capacity for "scheming" (covertly and strategically pursuing misaligned goals). We compare current research practice...

Vincent cites this paper, which looks worth reading. Reminds me of Clever Hans. If we want to believe something is "sentient," "conscious," or whatever other suitcase word has caught our fancy, we can make up experiments that yield fictitious evidence.

Against False Indulgences in LLM Alignment Unnecessary moral concern, and the false indulgences that drive it.

We should stop accepting vendors' framing of "reasoning" and "chain of thought," @vcarchidi.bsky.social argues in this provocative and well-argued essay. I really have nothing to add. We should remind ourselves that AI models are machines, not hunt for imaginary evidence that they're persons.

I might as well switch to rooting for Stoke if I have to follow a Championship club. What a crushing season.

Hope you all are able to stay safe, Ehud — thinking of you and your family.

Oh interesting. I’ve been struggling against gravity (as you term it). The agent having a project that is not really my thing has helped. But I bet fiction would help even more.

One of the best things about the AI discourse is that it's reminding people about Frank Herbert.

When people have responded to MeToo by saying "it was a different time," I think of my grandfather (born 1923) and father (born 1956) and how they interacted with the women around them. Treating people like human beings was not invented in the 21st century. There is no excuse.

With Friends Like These: Love and Friendship with AI Agents - Topoi This paper focuses on a central question for Human-AI interaction: Can you be friends with an AI agent? If not, why not? Some have argued that friendship with AI agents is impossible because software ...

Lott and Hasselberger make the same point in this excellent paper on chatbot fictionalism. (Which to be clear, argues that love and friendship with AI agents is impossible, the title is a bit sarcastic.)

The persona selection model A theory of why AI models act like humans

The persona that AI users interact with is *not* the model itself. That distinction keeps getting lost. I appreciate this explainer on persona selection that Anthropic has published.

There is a Darwin Award aspect to watching people respond to the AI hype!

Definitely staying at the Lutetia and eating at Grillades des Buenos Aires. Gelato at Amorino after.

When the weather sucks, we make Abuelita.

This has been on my list for a while, I should start it today.

Closing the AI benefits gap: Systems design for population health equity Artificial Intelligence (AI) is currently failing to live up to its potential. Its champions promise that it will make healthcare more effective, effi…

New paper out today in Public Health! 🎉
www.sciencedirect.com/science/arti...

AI's lackluster impact on pop health is not a technical issue - it's a design one. It has been given the wrong function, is being built on unsuitable infrastructure, & is too often too distant from patients/practitioners

This looks really useful, Jess -- I'm particularly interested in argument that frontline health care workers must be included in design and development. Sending you a DM now!

Yeah, in retrospect, "This technology can do *anything*!" is not super useful framing for adoption. When I teach in my industry, I spend a ton of time having people do concrete exercises. And I mostly pitch Claude and Gemini over ChatGPT, even though people walk in mostly w/ChatGPT experience.

In its current form, the discipline seems very susceptible to being taken over by the most efficient, grindy approaches, similar to what has already happened in the social sciences. I'd rather read weirder stuff produced less efficiently, too. And putatively "redundant" papers on big topics as well.

Tough to respond with nuance on Bluesky, suffice it to say I get where you're coming from. I don't see some of the "tool use" and sociality questions quite the same as you. But the relevant question, I think, is "what has academic philosophy become"?

Human × AI Finance Write a finance paper with AI. Get reviewed by AI. Top four presented at the Fink Center Conference at UCLA Anderson, April 24, 2026.

I'm strongly considered developing a paper for this. It feels like it would have a giant asterisk next to it compared to my other work. Exploring what not to do just so you can report on why not to do it.

I struggle with this. I've seen LLMs go from useless to very effective at editing. I use them daily for professional, non-academic writing. But for academic work, I feel like it would not be "my" writing to use them beyond grammar check and thesaurus stuff. Unsure if/where to push the envelope.

Fantastic thread on what to cover in a Philosophy of LLMs course.

Overheard while boarding: “Nothing like a transcontinental red-eye!”

Meanwhile, my spouse via text:

Yeah, it is good. Sometimes I struggle with STS / ethnography / sociology because it feels super thin. Goes on and on describing but is so afraid of passing judgment that it never even tells you clearly what is going on. This is the opposite of that. Conceptually crisp. Fast read.

Brian Eno's Ambient 1: Music for Airports once again proves it is indispensable for surviving airports.

I'm just finishing up Laura Voss's book, More than Machines?, which looks at the hype and reality of attributing animacy to robots. A lot of common incentives in both R&D and commercialization of robots circa 2021 and the current AI stuff, explaining some of the hype and anti behavior.

Derek Parfit's work had been lost by the 2370s unfortunately.

I know I’m old, because I’m anxiously checking the Warby Parker app to see when my new glasses ship. Everything is blurry.

Let’s do this for Helen, folks!

I am almost certainly going to quote this post in the paper I'm writing currently. Fair warning. :-)