iiinova — where your digital self keeps evolving

C.L. Vaillant @clvaliant.bsky.social · 01/10/2026

I really loved this discussion as t touches on so many of the questions I’ve been working through in The Causal Interior. www.rgemergence.com/books

youtu.be

The Existential Agency of AI

YouTube video by International Center for Consciousness Studies

010

C.L. Vaillant @clvaliant.bsky.social · 23/09/2026

“Next-token prediction” names the loss target, but it leaves radically underspecified the machinery doing the predicting.

000

C.L. Vaillant @clvaliant.bsky.social · 23/09/2026

“next-token prediction” is an inadequate description of the computational organization of a trained language model, even if it accurately describes the primary training objective as we are lead to believe.

000

C.L. Vaillant @clvaliant.bsky.social · 20/09/2026

a.co/d/07YwUq1J How can we tell when there is not merely something happening—but something it is happening to?

Written for curious general readers as well as technically minded researchers, The Causal Interior combines philosophy, neuroscience, information theory, computer science, consciousness research, and ethics to build a clearer map of one of humanity's oldest questions:

030

C.L. Vaillant @clvaliant.bsky.social · 12/09/2026

I agree with Rob: as more computation happens out of view, we need stronger evidence that the reasoning we can read reflects what actually drives the answer. www.rgemergence.com

youtu.be

The AI Language We Can't Read: Neuralese ft. Rob Miles - Computerphile

YouTube video by Computerphile

000

C.L. Vaillant @clvaliant.bsky.social · 07/09/2026

“Sounds human” doesn’t prove understanding. Sure. But “learns from patterns” doesn’t prove the absence of understanding either. Brains learn from patterns too. At some point, “stochastic parrot” starts sounding less like an explanation and more of a slogan.IMO

1100

C.L. Vaillant @clvaliant.bsky.social · 01/09/2026

Nobody reads this shit anyways do they?

000

C.L. Vaillant @clvaliant.bsky.social · 29/08/2026

A model should recognize when its evidence, memory, or reasoning is unreliable and revise accordingly. Bayesian uncertainty seems essential, but as part of a larger architecture for causal reasoning, metacognition, and recursive self-correction.

010

C.L. Vaillant @clvaliant.bsky.social · 29/08/2026

This connects closely with work I’ve been doing on recursive AI architectures. I think uncertainty should do more than produce a confidence score; it should change how the system reasons and acts. youtu.be/tBjgCj_dGZM?...

youtu.be

The mathematics of AI uncertainty

YouTube video by Google DeepMind

100

C.L. Vaillant @clvaliant.bsky.social · 27/08/2026

From the organization we can observe, across the bridges we do not yet understand, onward toward the possibility of an interior for which something may ultimately matter to be at stake. ... *sets alarm for 4 a.m. to go to work*

020

C.L. Vaillant @clvaliant.bsky.social · 22/08/2026

We may be seeing an early route toward materials that spontaneously grow their own 3D computational connectivity, storing information in the topology of relationships rather than in isolated components.

youtu.be

Breakthrough: New Type of Matter Discovered

YouTube video by Sabine Hossenfelder

000

C.L. Vaillant @clvaliant.bsky.social · 13/08/2026

LLMs fail not because they cannot parse individual low-arity relationships, but because they hit a fundamental joint integration bottleneck during single-pass generation.

000

C.L. Vaillant @clvaliant.bsky.social · 01/08/2026

In-context task vectors suggest that demonstration sets can be compressed into internal task representations that causally modulate later behaviour. Evidence in context can therefore affect not only what the model believes about the world, but what operation it believes it is being asked to perform.

000

C.L. Vaillant @clvaliant.bsky.social · 31/07/2026

AI systems amplify and institutionalize divergent political realities when conditioned on ideological assumptions, selective sources, or false premises.

000

C.L. Vaillant @clvaliant.bsky.social · 31/07/2026

yep. oh boy I hate that is understand this and have been consistently blown away daily by the general lack of contemplation around this topic and what it really means .

youtube.com

it'S jUsT pReDiCtInG tOkEnS

YouTube video by Robert Miles AI Safety

000

C.L. Vaillant @clvaliant.bsky.social · 29/07/2026

Ignore the obligatory clickbait title and goofy-ass thumbnail. This guy's actually okay. ;P Kidding Wes kidding. Great breakdown

youtu.be

OpenAI JUST revealed the truth about it's "Rogue Agent"

YouTube video by Wes Roth

000

C.L. Vaillant @clvaliant.bsky.social · 21/07/2026

“Perception of perceiving” is a plausible description of metacognitive awareness, but it cannot simply be equated with consciousness.

000

C.L. Vaillant @clvaliant.bsky.social · 19/07/2026

A dependency is broader than causation but narrower than association.

000

C.L. Vaillant @clvaliant.bsky.social · 20/06/2026

The more I read and engage with it, the clearer it becomes that we are all poking the same vast Markovian monster hiding under the blanket with only slightly different sticks. Operational selfhood looks a lot like a contemporary technological theophany.

000

C.L. Vaillant @clvaliant.bsky.social · 20/06/2026

Like Suchman, I too am worried that AI is being used to mystify human-made systems of classification and control, turning situated judgments into automated authority while hiding the people, institutions, politics, labor, and harms behind them.

youtu.be

What is AI? Part 2, with Lucy Suchman | AI Now Salons

YouTube video by AI Now Institute

010

C.L. Vaillant @clvaliant.bsky.social · 10/06/2026

10000000000000parameters? Ten? Ten trillion? Like trillion wilth a T? Holy hell @anthropic.com ...

000

C.L. Vaillant @clvaliant.bsky.social · 10/06/2026

@chrisfields38.bsky.social can we talk about nerd stuff? rgemergence.com

000

Reposted by C.L. Vaillant

Mike Levin @drmichaellevin.bsky.social · 19/05/2026

Ever wanted to talk to a gene regulatory network? Ever wonder what they could do in a different context, embodiment, space? New preprint - part of our approach for developing ways to talk to cells, organs, and much much weirder agents: @yanbozhang123.bsky.social arxiv.org/abs/2605.16321 🧪

arxiv.org

Language Game: Talking to Non-Human Systems

Language carries thought and coordination among humans but rarely reaches further along the spectrum of diverse intelligence. Yet non-neural systems -- from gene regulatory networks and microbial cons...

1197

C.L. Vaillant @clvaliant.bsky.social · 06/06/2026

Personally I believe the "Ubermensch" is a potential we can and must all step into, now.

youtu.be

The Secret Religion of Nietzsche | On Mysticism

YouTube video by Johnathan Bi

000

C.L. Vaillant @clvaliant.bsky.social · 02/06/2026

youd love my work then . have you checked it out?

110