Paul Harland
banner
pabloredux.bsky.social
Paul Harland
@pabloredux.bsky.social
Evolution: it's nothing personal. (Sorry if I ask too many questions... Share nicely!)

I'm here because you're here.

http://pabloredux.wordpress.com/

(Profile photo: a faint shadow of me cast across a weathered lichen-marked stone wall by the sea.)
Typos?
November 19, 2025 at 12:48 PM
I think she asked good, reasonable, questions that Trump couldn't satisfactorily answer, but his tone will have derived in part from the preceding Khashoggi question, not just the Epstein one.
November 19, 2025 at 12:21 PM
A fine take!
November 19, 2025 at 8:50 AM
I think that's only partially true? Protests by activists can be representative or unrepresentative or an attempt to become representative.
November 19, 2025 at 8:49 AM
It might've been set up by ABC's previous question?
November 19, 2025 at 1:01 AM
😄
November 18, 2025 at 6:11 PM
Till you change your mind.
November 18, 2025 at 6:11 PM
So I haven't happened on anything that explores the gaps - my reading is sporadic - but you're probably right, Yohan: bsky.app/profile/dryo... There is stuff on the strangeness of confronting LLMs, but mostly too concrete or too abstract. David's piece helps delineate part of the conceptual puzzle.
I've tried googling new papers but nothing seems to confront the strangeness of the present moment.
November 18, 2025 at 5:01 PM
I'm sure there are papers that engage in a much wider discussion of what goes on within any instance of an LLM following a prompt - 'cos this is quite a specific kind of "self-attention" - but I'm just trying to remember stuff that caught my eye: bsky.app/profile/pape...
[23/30] 198 Likes, 1 Comments, 5 Posts
2507.16003, cs․CL | cs․LG, 21 Jul 2025

🆕Learning without training: The implicit dynamics of in-context learning

Benoit Dherin, Michael Munn, Hanna Mazzawi, Michael Wunder, Javier Gonzalvo
November 18, 2025 at 4:47 PM
Another perspective on self-presence would be to see what kind of "self" (or selves, or roles) an LLM performs compared to what it "possesses". Again, I'm assuming there are necessary processes missing, but presumably a chat session still constructs a minimal - and relatable - reflexive structure.
November 18, 2025 at 4:27 PM
Also caught my eye in the same journal: bsky.app/profile/lars... Good to pair with Chalmers consc.net/papers/quali... (thx Oisín Mac Fhearaí: www.linkedin.com/feed/update/...) on whether "organization realized by the brain at the neural level might in principle be realized by a silicon system".
Do bees see nothing? Primates with damage to the visual cortex have no conscious visual experience, but display adaptive visual behaviour (a phenomenon called blindsight). Here I team up with two primate researchers to ask if bee vision could be similar to blindsight. www.cell.com/trends/cogni...
November 18, 2025 at 4:10 PM
What other gaps matter? "What is the 'point' of rendering, if any? Physical simulation is sufficient to solve many tasks taken to involve mental imagery..." www.cell.com/trends/cogni... Should "agent representations in the mind ... be seen as part of the physical simulation or... separate from it"?
Physics versus graphics as an organizing dichotomy in cognition
People build world models that simulate the dynamics of the real world. They do so in engineered systems for the purposes of scientific understanding or recreation, as well as in intuitive reasoning t...
www.cell.com
November 18, 2025 at 3:55 PM
Heaven is for poor people. Wouldn't catch me dead in there.
November 18, 2025 at 3:27 PM
Although I'm not a total fan of phonics. Anecdotally it's possible to score excellently at the phonics "game", love books, but not read fluently. This is probably only a small minority of readers but seems possible to me different readers work around their strengths and weaknesses in different ways.
November 18, 2025 at 1:58 PM
I've been trying to needle @davidgunkel.bsky.social to look more into the gaps between LLMs and our minds bsky.app/profile/pabl... but I'm an amateur.
In relation to @davidgunkel.bsky.social's 'Différance engine: large language models & poststructuralism' link.springer.com/article/10.1... and pabloredux.wordpress.com/2014/10/09/a... "We can train our bodies to do as they’re told, but oddly, if we do it too well, they stop asking us what to do."
November 18, 2025 at 1:42 PM
Happened a few months ago in UK, red crosses on white backgrounds. Avoid.
November 18, 2025 at 12:47 PM
Sad tales of marginalisation in the overdeveloped world: bsky.app/profile/jim....
November 18, 2025 at 12:05 PM