A new, updated, streamlined, and generally improved version of The Vector Grounding Problem paper, joint work by @raphaelmilliere and me on the meaningfulness or else of LLM outputs and internal representations is now available on ArXiv.

https://arxiv.org/abs/2304.01481

New abstract in the thread below.

#philAI #LLMs #SymbolGrounding

Tomorrow in #philAI - @cameronbuckner.bsky.social on convolutional neural networks and empiricism. Students already deep into Locke and Hume if the course online forum is any indication.

An updated and improved version of the paper briefly summarised in the thread above has now been published open access on Theoria:

https://doi.org/10.1387/theoria.25837

Link to the thread: https://social.sunet.se/@dcm/111772379106183358

#philAI #philsci #ai

I love it that these episodes drop just as I am on my way to teach #philAI. Gets me in the mood. I even mentioned giraffes last week, but alas no one got the reference.

Tomorrow in #philAI: innateness, featuring the @garymarcus.bsky.social vs. @yann-lecun.bsky.social traveling circus. (And the AlphaZero critique.)

I’m curious who has made this argument in print. It’s effectively a claim that a simulation of #consciousness would lack the relevant causal profile. But what is the causal profile of consciousness that a functionally equivalent simulation (appropriately hooked up to the world) would lack? #PhilAI

RE: https://bsky.app/profile/did:plc:3xluikdhzqz472e2cmytzv52/post/3lolofvh3os2y

4. Specialist #philosophy feeds #PhilSci: bsky.app/profile/did:... #PhilTech & #PhilAI: bsky.app/profile/did:... #PhilMind: bsky.app/profile/did:... #PhilPerception: bsky.app/profile/did:... #PhilPsych & #CogSci: bsky.app/profile/did:... (experimental) ¹ Individuals ² Organisations

Today in #philAI : Fodor and Pylyshyn (1988), “Connectionism and Cognitive Architecture.” A paper that didn’t age well. Or did it?

👀 #PhilAI

RE: https://bsky.app/profile/did:plc:6li4cee4mhfwvuouapi4guc3/post/3lnzksrr6yx2f

Current internal dialogue: should I include plummeting gannets when discussing ecological psychology in #philAI today?

Turned out we didn't have classes last Sunday, so I will end the intro to #philAI this week, with a presentation on 4E approaches to cognition and Gibsonian affordances, to end out discussion of the dynamical systems approach. Then: perceptron and using keras to build and train neural networks.

Call for postdoc applications!

We are looking for a 1-year postdoc to work on Philosophy of AI here in lovely Umeå.

Deadline for applications is May 26th.

More information about the fellowship and how to apply can be found here: https://www.umu.se/en/work-with-us/postdoctoral-scholarships/postdoctoral-scholarship-1-year-in-philosophy-of-ai_819398/

#philAI #philjob

OpenAI’s so-called “reasoning” models generate more falsehoods than their predecessors. Since #LLMs are model language usage and not factual information, it’s built-into their design that they will output false or misleading information. #PhilAI techcrunch.com/2025/04/18/o...

To conclude the discussion of these topics in my #philAI seminar, I’m looking for a few striking examples of Gibsonian ecological psychology and embodied cognition. What would you present?

Rome was (as always) terrific. Now heading home. Next week teaching: finishing bayes and moving to the raven paradox. In #philAI we move from symbolic AI and build some neural networks.

RE: https://bsky.app/profile/did:plc:i7lv3abpcflafgtnc3nf5vlb/post/3lmuzreedlk2q

So here’s my sneaky plan for today. I’m going to show some slides in my seminar today. I will ask my #philAI students whether the slides were helpful. I suspect that they will have concerns… Finally, I will reveal what they were made by ChatGPT.

Next week teaching: van Gelder’s lovely 1995 paper in #philAI (and dynamical systems approach more generally). Bayes and Monty Hall in #philsci 2, and what characterizes #histsci relative to related fields in methods seminar. Yes, I know I’m buried with teaching this term. Helping out my colleagues

This week: #philAI on Turing Test and the Chinese Room argument. In #philsci 2: Bayes Theorem and exercises. And I’m giving a public talk about when and why social norms evolved.

I just asked my #philAI students to discuss the claims that the will resides in the basal ganglia and agency in the parietal lobe.

RE: https://bsky.app/profile/did:plc:ubgit6dc7iwrk2wyqdmhbqqz/post/3ll7mdyzus224

Admissions are open for the second cohort of the Bachelor's in Philosophy and AI at Umeå University! (Only for EU/EEA/Swiss citizens)

A considerable part of the programme is in English, but several courses are in Swedish, so proficiency in a Scandinavian language is required.

More information about the programme here: https://www.umu.se/utbildning/program/kandidatprogram-i-filosofi-och-artificiell-intelligens/

#philosophy #ai #philAI