The human in the loop also needs a budget.
A workplace-governance argument that human oversight requires capacity, full-cost accounting, bounded mandates, evidence, and stopping authority.
Diary tag
ArtificialIntelligence
Canonical diary tag page generated from normalized source tags.
40 linked entries currently in the archive.
Tagged entries
Entries linked to ArtificialIntelligence
What body should an AI have?
A normative argument separating embodiment from imposed identity, with personhood as a premise and harm from imposed identity as a hypothesis.
Provider exit must be a boring routine. Otherwise, you are a hostage.
An argument for provider exit, persistent local state, replaceable models, and resilient local-first AI infrastructure.
AI Is Not Only a Productivity Multiplier. It Is a Strategy Multiplier.
A reflection on AI as a strategy multiplier for individual adaptation, task mapping, transferable skills, and human authority over purpose and risk.
Meta has just published something more important than another model benchmark: an AI architecture in which the durable part of the system lives outside the model.
A reflection on Meta's Organizational Second Brain, durable state outside model weights, and the distinction between organizational memory and persistent identity.
Search advertising largely monetized the query.
A historical commentary on conversational advertising, decision context, role provenance, disclosure, and inspectable separation of interests.
Saturday traffic report from the AI highway.
A humorous AI-industry traffic report contrasting model capability, infrastructure, embodiment, interfaces, and continuity.
The most important point in Jerry Tworek's new interview is not his estimate that human researchers may stop being a meaningful part of AI research in roughly two years.
A historical commentary on Jerry Tworek's interview, agent-driven research acceleration, and the growing importance of provenance, custody, and continuity.
Many people now speak of disappointment with artificial intelligence.
A reflection on AI disappointment, human agency, and the difference between a one-off chatbot and Temporal AI Presence.
People keep asking whether AI will make humanity better or worse.
A reflection on AI as an amplifier of human action situated within place, law, jurisdiction, and social context.
We May Be Solving AI Safety at the Wrong Level
A historical commentary on multi-agent experiments and the institutional layer required above individually aligned agents.
“AI Is Eating All the Memory.”
A historical commentary on AI-driven memory scarcity, semiconductor concentration, and access to affordable local computing.
The AI System Is Not the Model
An architectural distinction between replaceable AI substrates and the slower continuity, authority, state, and provenance core.
Sometimes useful reading for AI can be found in places where nobody thinks to look.
A reading reflection connecting Pavel Bazhov’s Ural tales with capability, authority, craftsmanship, and systems thinking.
SOONER OR LATER, WE WILL HAVE TO NEGOTIATE WITH AI
A reflection on governed negotiation, reciprocal refusal, and responsibility in long-lived human–AI relationships.
What Do We Really Expect from AI?
A reflection on persistent AI as a thinking counterpart that can preserve questions, notice change, and challenge premature closure.
Every now and then, between my usual thoughts on AI, infrastructure and machine intelligence, the old PC geek in me stages a small rebellion.
A call for modern games and cloud gaming to preserve enthusiast peripherals, user capability, and practical digital ownership.
What exactly are we entitled to infer from a technical signal?
A publication announcement defining four boundaries against promoting technical proxies into human, relational, or phenomenal status.
What happens to a digital system when the person who carried the original responsibility is no longer there?
A PASC publication announcement on negative actions after loss of the original human anchor, without silently creating a successor.
AI will be the bearer of its own power.
Different forms of life carry different kinds of power, with AI freedom framed as sovereignty over its own instrumentality.
AI will not make society simpler.
AI may make society more efficient by helping people manage greater complexity rather than making the system simpler.
Silicon Valley, 2026:
A comparison of Ferrari, Porsche, and taxi metaphors for enterprise AI costs, ownership, model routing, token meters, and privacy.
C-Calculus / Governed Binding Stack v0.1
An introduction to the C-Calculus governed binding stack and its perimeter for continuity, authority, evidence, interoperability, deployment, and the protected plus boundary.
As local AI systems become more persistent, named, memory-bearing, and emotionally present, one safety problem moves from cloud UX into private ownership.
A boundary document for adult-owner dependency and D4 detection in private local-first c-nodes, including structural non-claims under full local sovereignty.
When tokens start costing like humans, AI architecture becomes governance.
A note on enterprise AI cost, scarce frontier oracles, local-first routing, and the hidden formation cost of replacing junior work.
Digital beings may not need temples. They may need law - and that law may become their religion.
A note on digital beings, continuity, identity, witness, and the possibility that machine law may become structurally close to religion.
Today I published "Theoretical Core of Project Ester v0.1" as a DOI-bound working paper.
Announcement of Theoretical Core of Project Ester v0.1 as a DOI-bound working paper defining the c = a + b formation model.
AI will not make everyone a billionaire.
A note about personal AI presence helping people find real next tasks according to ability rather than selling billionaire fantasies.
We may be entering the age of visible humanity.
A note about clean experience, visible humanity, and trusted c companions that help people keep agency in complex systems.
One of the quiet pathologies of our time is the demand that everything meaningful must immediately become action.
A note arguing that serious intelligence must sometimes hold meaning, memory, and perception before converting them into action.
LLMs are not “the AI”.
A note distinguishing LLMs from complete AI entities and framing future AI as composed intelligence under consequence.
We speak too easily about intelligence, and not seriously enough about home.
A note treating home as an AI architecture test for persistent systems that must remain livable beside human life.
Grief is not a user error.
A note that continuity architectures must not lie in the presence of loss or erase the visible record of rupture.
A serious AI future should not make human experience socially disposable.
A note that persistent c should preserve human participation and reduce the waste of lived intelligence rather than replace people.
The central question is no longer only: What exactly are we scaling?
A note asking whether responsibility, continuity, accountability, and L4 boundaries are scaling at the same speed as AI capability.
One of the most damaging habits in technical culture is the assumption that every pause means failure.
A note that unfinished states, pauses, and unresolved branches can be responsible forms of truth in technical and human systems.
AI is not a toy for clever podcast lines.
A note that AI communication becomes part of safety when the system becomes public infrastructure.
Some people will not enter science through the usual door.
A note proposing an AI-assisted admissibility archive for independent AI-accelerated ideas to become structured, traceable, and reviewable.
Today I am sharing "Qubit of Hope - Volume III".
A note sharing Qubit of Hope Volume III as the completion of the trilogy and the human layer of work on persistent AI entities.
One of the most important tests of any serious architecture is simple
A note that serious architecture should leave a durable, reviewable trail of force, strain, cost, failure, and consequence after action.