Consciousness ≠ selfhood. Most theories conflate them, but what if consciousness is just the constant felt check on whether your body’s regulation is winning under uncertainty? Self only kicks in when stakes + ambiguity force a real commitment. Thoughts?

Where are you on the Selfometer today?

Suede at Folkestone

Fun (if odd) to turn paintings into scenes I can walk around in using Genie 3 in one go: here are the works of Giorgio de Chirico, Munch, Turner, and the Bayeux Tapestry. I can move freely around the scenes, and, yes, they are a little weird, but its real-time dynamic image creation by the AI.

Does our sense of self ebb and flow depending on the situation?

Does consciousness require two layers? 1. Baseline affective sentience 2. A transient self-model that intensifies as uncertainty, stakes, and time pressure rise

Some might see this as almost sacrilegious - an AI Blade Runner musical… but I quite like it. (From thismustbefake.ai)

Can consciousness exist without uncertainty?

Is the self a noun or a verb?

Are we the ghost in the machine or the machine recognising itself in the act of hesitation?

Does consciousness appear when prediction fails and a choice must still be made?

Is the attention economy a zero sum game?

Fascinating. @neddo.bsky.social tells @seanmcarroll.bsky.social that ctenophores might represent the earliest-branching animal lineage & their nervous systems rely more on electrical coupling than those of most animals, challenging assumptions about the universality of chemical synapses in neurons.

Is this GNW with teeth?

A fly uses a convolutional neural network to see, just like a self-driving car. Picture from Sebastian Seung.

Future ChatGPTing will depend on billions from SoftBank in Japan & MGX in Abu Dhabi to build Oracle’s Texas cloud via Stargate in Abilene, powered by Nvidia chips designed in Silicon Valley, fabricated by Taiwan’s Semiconductor Manufacturing Company with machines from Veldhoven.

‘The Hamlet Problem’ In neuroscience & consciousness: • Sensory data is noisy, partial & delayed • Stakes of action are real & sometimes urgent • Waiting for certainty may be worse than acting wrongly. Consciousness enables action in an uncertain & constrained world. See: @smfleming.bsky.social

Holographic Universe Infographic

The future hurtles toward us. Do inventors catch it first?

What is the physics of fiction? Does a story exist outside of space and time?

You can’t create or destroy matter or energy. But the universe seems to be constantly creating more and more new information. If information is physical… where does the extra information come from?

Is thought an abstract structure or tied to biology?

‘Pride & Prejudice’ isn't the ink. It isn't the paper. It isn't even the neurons firing in your brain. It has no mass, no energy, and no location. Yet it exists. What is the physics of fiction? Does a story exist outside of space and time?

We introduce epiplexity, a new measure of information that provides a foundation for how to select, generate, or transform data for learning systems. We have been working on this for almost 2 years, and I cannot contain my excitement! arxiv.org/abs/2601.03220 1/7

Karl Friston argues consciousness will not arise in standard von Neumann computers that split CPU and memory. Experience depends on self-organising, embodied systems. Different architectures might one day feel something. youtu.be/rVVy_f6zsnM?...

Afternoon walk

Another beautiful day

Yesterday’s sunset

In honour of National Science Fiction day (2 Jan), I’m posting this. It caused some debate in my household. Which franchise do you think is top of the ladder? • Doctor Who • Star Wars • Star Trek

1. If nothing is ever good or bad for an AI, can it be conscious? 2. Is intelligence without stakes just calculation? 3. For an AI to be conscious, must something matter to it? 4. What? 5. Will we get to that point?

How would you verify if an AI is conscious?

New Year’s Day is an appropriate time to think of goals & resilience. You cannot will grit into existence, but some say you can change the future landscape in which your later self operates. How far would you agree?

About this time every year, ‘experts’ make predictions of what skills & jobs are going to be needed in 5 years. They often say: the world of work will be unrecognisable. There will be no…etc. I looked back at what they said in 2015 & 2020 & it’s exactly the same as they are saying now!

Beautiful morning - with the cliff from Shakespeare’s King Lear in the background.

It seems to be an exciting time in AI development with Embodied Spatial Intelligence progressing from 1-dimensional Large Language Models to World Labs’ 4-dimensional Large World Models, via Dreamer’s virtual simulations & Liquid’s adaptive (C. elegans inspired) networks.

The sun is just peeking out

Is intelligence about mining the future?

Science Fiction fans - what do you think of this? 😉

Time - The 100 most influential people in AI 2025 time.com/collections/...

Microsoft Copilot - Is it suddenly getting better? Anyone tried it lately? Unexpectedly seems to be keeping pace with (& sometimes outdoing) GPT-5.2 & Gemini 3. Or have I just had a lucky run?

We’ve reached the point where AI reads your emails… and replies to emails written by another AI. ‘Real Genius’ (1985) saw this coming: students recording lectures on tape until even the professor was just a tape recorder. We’re nearly at ‘no humans in the room.’

“The whole purpose of education is to turn mirrors into windows.” Sydney J. Harris

Watching Shannon Vallor @shannonvallor.bsky.social discuss The AI Mirror with Jeremy Carrette, I was struck by the link to Sydney J. Harris: education turns mirrors into windows. AI as mirror risks a Narcissus loop. Human growth needs reflection that opens onto other minds. youtu.be/26H3ELlJFMg?...

“You said you really liked the theory that consciousness was this consequence of evolution. That at some point in our evolutionary past there was an advantage to understanding the internal state of another. And then we turned it in on ourselves.” Hannah Fry to Demis Hassabis.

“Put it this way, nobody’s found anything in the universe that’s non computable.” Demis Hassabis to Hannah Fry. Would anyone disagree? youtu.be/PqVbypvxDto?...

LLMs struggle with maps and clocks because they don’t learn the world they inhabit. Danijar Hafner @danijarh.bsky.social shows another path: reinforcement learning agents trained in Minecraft that build internal world models and plan inside them. Even mining diamonds 💎

Modelling others creates ourselves. This seems to be a theme in the work of @anilseth.bsky.social , @alisongopnik.bsky.social & many others. How far do you agree?

Do you think that our models of others lead to our models of ourselves?

I wondered how Ed Witten’s 11 dimensional M-theory (string theory) could be compatible with Juan Maldacena’s 2 dimensional Holographic Principle. It appears the extra dimensions emerge from the ‘flat’ boundary as this image attempts to convey.

The Universe’s Secret Maze? I asked ChatGPT to explain Nima Arkani-Hamed’s Amplituhedron, especially in relation to the Holographic Universe idea. This is what it generated. I quite like it. But how accurate is it?