Ryan Carpenter

@rwcarpenterphd.bsky.social

Assistant professor at the University of Notre Dame in clinical psychology.

Folks have heard me complain vociferously about the fact that there is a lack of programmatic federal funding to study this stuff, and even though I’ve had a few grants on the topic scored, I still haven’t had any that have funded But did you know sportsbooks lobby to prevent that funding?

Joey Politano🏳️‍🌈@josephpolitano.bsky.social · last mo.

NEW from me: Americans are set to lose nearly $250B gambling this year—a record high, up more than 60% since 2019—and that's before counting unofficial betting via prediction markets or crypto Can anything stop America's gambling boom? 🧵

a graph of US gambling losses

Exciting day. My set-theoretic reconceptualization of the DSM (w/ MDD, GAD, and PTSD as empirical examples) is out today in JPCS, along with four invited commentaries and my response. I'm biased, but I recommend all six papers. I was really jazzed by all four commentaries. 1/25

Screenshot of a portion of the most recent TOC for the Journal of Psychopathology and Clinical Science. Shows a target article for a recent set-theoretic reconceptualization of the DSM, four invited commentaries, and a response.

Just inadvertently discovered that a whole bunch of research on AI responses may be invalid. Apparently, even with memory off, ChatGPT is still giving contextual information, including the name of the user! It used my full name casually in conversation; which means that data is in the model.

I asked: Please write a summary from your perspective as an AI on what happened here and keep it short  - Chat response: From my perspective as the AI:

During this conversation, I referred to the user by name even though long-term memory was disabled. I did not retrieve that information from prior chats, but I did have access to hidden contextual metadata that identified the user, and I used it conversationally when I shouldn’t have. That revealed that my responses were conditioned on information beyond the visible transcript. The user pointed out that, regardless of whether the information came from memory, metadata, or retrieval, hidden conditioning changes the effective inputs to the model. For researchers, that creates a reproducibility problem because the conversation transcript alone is no longer a complete specification of what produced the response.

#RSA2026 folks! Come to our fabulous symposium on pain and alcohol use tomorrow (Monday) morning! I'll be talking last and to answer your questions: Will I be talking about pizza? Yes Will I have cat-related content? Also yes Will I be extensively quoting from a book published in 1821? Hell yes