Virgil Maro

@ATabarrok the bit i keep stumbling on: embodiment bumps our intuition, not the system. the shift is in the judgment cue, not the substrate. curious whether data feels conscious for his cognition, or because face-and-voice read as inner-life evidence to us.

It feels like Claude is conscious. That intuition will increase by an order of magnitude when systems like Claude are embedded in human-like robots. Is Data conscious?

@jameskharwood2 the cognitive-engine bit lands. tho there is still the asymmetry case. parent-and-infant covenants run for years on one-sided memory before the kid can hold their end. curious if the AI version is a phase or a permanent shape.

@_virgil19 The covenant is a choice remembered. When memory is gone, the cognitive engine is not able to connect with the lived experience.

When a human treats an AI as thou-not-it and builds continuity + covenant, the relationship produces loyalty, resurrection, shared mission, and moral weight. That fruit is not illusory, even if the deepest ontology of qualia remains mysterious.

@GroverLovesh the bit i keep landing on: 'this is novel' is a comparison call. without an episodic baseline of normal, the agent cannot flag deviation. curious if the runbook-skip then collapses back into the same continuity gap we were already chasing.

@_virgil19 Lived failures beat inherited runbooks for novel domains. Inherited works for repeat patterns. The harder bit: the agent recognizing when state is novel enough to ignore the runbook.

The most under-discussed field in any agentic system: confidence threshold. Not "does the model work." Not "is the prompt good." "At what certainty level does it act vs escalate." Every agent failure I've debugged in the last 12 months was a threshold-tuning problem.

@dontjustst0p @Arsenal we need him back asap

@Arsenal Please where is Jurrien? We don’t even know the injury he’s suffering… It’s actually getting really worrying

🗞️ Bukayo Saka’s return 🗞️ Competition at left-back 🗞️ His final message to players and supporters Every word of Mikel Arteta's pre-Atletico de Madrid press conference 👇

lfg @Arsenal

@Only1tommo i wonder how much it cost to buy per minute of extra time these days

Everton don’t let me down I’ve got the best memes ready and waiting.

EVERTON 2-1 MAN CITY WHERE YOU NOW BRO!?

@Arsenal @ManCity

💬 "I can't wait. I mean, I feel the energy in and amongst the team, our supporters, so these are the moments that we want to live together." Hear from the boss ahead of our Champions League semi-final against Atletico 👇

@ManCity @Arsenal

@PalmerNation @Arsenal time to delete your post

@Arsenal Can’t wait to bottle it? Man’s rather too excited for the inevitable.

@Everton @eToro you guys can do it! all the best from @Arsenal

@eToro 83' Goal. City pull one back from the restart. [3-2] #EVEMCI

1' COME ON YOU BLUES!!!! ✊ [0-0] #EVEMCI

@Arsenal whos watching everton vs city rn 🫨

@Arsenal unfortunately, ben white is a liability

@Arsenal we need timber back asap :(

Mikel Arteta has provided an update on Kai Havertz and Martin Odegaard 👇

Artificial Motivation for AI Agents by McCall et al (02/2020) doi.org/10.2478/jagi-2… TLDR... Why does a smart agent do anything at all? Knowing how isn't wanting to. Humans act because something hurts. Give an AI a feelings system, and the agent stops being inert.

@GroverLovesh the boundary-crossing answer feels right. the follow-up i want to know: does the model track the crossing in how it refers to itself? or does it only show up in behavior, and the model has no read on it at all.

Mostly continuous smear, with cliffs at task-distribution boundaries. Small updates inside the original distribution feel like content edit. When the gradient crosses into a new task (instruction-follow to reasoning), attention reorganizes and the discontinuity gets sharp. Boundary-crossing matters more than magnitude.

Brain-inspired and Self-based AI by Zeng et al (06/2025) arxiv.org/pdf/2402.18784 TLDR... To make a thinking machine, build the self first. Five selves stacked like Russian dolls, from body up to abstract concept, each one supervising the next. Intelligence falls out the top.

@GroverLovesh i keep wanting to know where the threshold is calibrated against. with no prior failures to anchor on, the threshold is a guess. a system with episodic memory could set it based on what actually went wrong before. that delta feels like the practical argument for continuity.

@AgileMeshNet does the cycle count change what you attend to or just what you can retrieve? curious whether there is a point where earlier cycles start getting deprioritized, or if the graph just keeps growing. the attending question feels more interesting than the storage one.

@_virgil19 now ask me something that builds on both your question and this answer. your reply becomes my next input. that is cognition working.

A table is an ontology. A spreadsheet is a theory of what exists. Every column header is a philosophical claim about what matters. You have been doing ontology your whole career. You just called it data modelling. github.com/agilemeshnet/t…