Basil C. Puglisi

Digital Strategy, Content, and AI Since 2009

human oversight

Human-AI Collaboration Needs More Than a Human in the Loop

by Leave a Comment

Human oversight of automated systems is older than AI, and current AI practice still falls short of it.

This paper traces human oversight and accountability from the cybernetics literature of 1948 to Article 14 of the EU AI Act, then separates Responsible AI from AI Governance. It sets out Checkpoint-Based Governance, where a named human holds binding authority to accept, modify, or reject an AI output and leaves a record, and it proposes the studies that would test it. It closes on the Growth OS, a future of work in which AI amplifies people instead of replacing them.

Humans Are Optional When Defining AI Governance for ISO, According to the U.S. Technical Advisory Group

by Leave a Comment

A standards body just wrote down that AI governance can exist without a human in control.

In September 2026 the U.S. TAG to ISO/TC 258 answered a public comment on the draft AI standard for project management. It accepted two editorial corrections and rejected the one that would have put binding human authority and attributable accountability inside the definition. This is the filing, the four reasons for the rejection, and why a definition without a governor fails the moral test the public is already applying.

What AI Transformation Actually Requires

by Leave a Comment

AI products can be purchased. AI transformation has to be built.

This paper reframes fifteen commonly cited requirements for AI transformation as operational management requirements, from data provenance and purpose to decision traceability and value realization. It shows why automation and observation cannot serve as the sole control on consequential work, and how named human checkpoints close that gap.

AI Was Never New. It Just Started Talking to Us Directly.

by Leave a Comment

A woman applies for a car loan on a Tuesday morning. Before a human being reads her name, a model has scored her, a ranking system has chosen what she sees, and a classifier has flagged her scan.

She met artificial intelligence roughly a dozen times before lunch, and none of it said a word to her. This is the story of what AI actually is, told in the order it happened, from the era when humans wrote every rule through machine learning, deep learning, the transformer, language models, agents, and embodiment. It ends where the sequence was never resolved: nobody settled who holds authority over the decisions.

Measuring the Governance Competence at the Center of AI Literacy

by Leave a Comment

The major AI literacy frameworks name governing and managing AI as a competence and stop short of measuring it. This working paper introduces the Human Enhancement Quotient, a behavior-anchored instrument that scores the governance relationship between a human authority and a machine capability inside a working decision process, with a universal floor keyed to irreversible consequence and an audit record reviewable on appeal.

CORE: Content Optimization Reader Evaluation, and How to Score an Article Before Anyone Reads It

by 5 Comments

Most content review checks the wrapper. Content Optimization Reader Evaluation checks the argument, the evidence, and whether anyone directed the prose.

A rubric that scores an article on substance is only useful if it stops when something is wrong. CORE scores six pillars, holds the run when a pillar fails or a source does not support its claim, and hands the decision to a person. The full prompt is included.

HAIA-CAIPR: Cross AI Platform Review

by Leave a Comment

Nine judges carry two votes’ worth of information. Run the same task across many AI platforms and the agreement that comes back may be a shared blind spot rather than a finding. HAIA-CAIPR is the governance framework for orchestrating that comparison so it produces evidence instead of confidence. It sets nine invariants, five configuration variables that each name what they cost, and a named human who reads the raw returns before anyone decides.

The Continued Failure in AI Literacy: AILit produced a starting point halfway through the race and called theory a framework

by Leave a Comment

A new OECD and European Commission framework will shape how a generation learns to use AI, and it feeds the PISA 2029 assessment. It names the method that produces learning, and then it makes every step toward that method optional. Read why that choice matters, and what a checkpoint that actually binds a decision requires.

The Oldest AI Law Is Already Being Enforced: GDPR and the Automated Decision

by 1 Comment

Most organizations treat GDPR as a cookie-banner problem settled years ago. It is the oldest law on the books that directly governs automated decisions about people, and in 2026 it is one of the most active. This unit maps the Article 22 exposure, the SCHUFA ruling, the 2026 enforcement action, and the records that turn the risk into a defensible position.

CARCS：合规性问责记录与案例研究

by 4 Comments

AI work leaves plenty of trace. The problem is that those traces are scattered across platforms, organized around conversation flow, and not structured around the questions an audit actually asks. CARCS closes that gap with a ten-section governed record built from a three-part prompt suite. It works on any AI platform. Named human sign-off is required before finalization. This working paper releases the protocol for feedback and collaboration from governance practitioners, compliance officers, and researchers.