Authorship And Credits

Release candidate. The unit videos are added in the first release.

Made by several AI systems working with a person, Frank C. Langbein, who reviews the work, gives advice, and releases and maintains it.

This course teaches learners to state who made what and to keep "I generated this" separate from "I understand and can defend this". It would have no standing to do that if its own credits were vague. So this page is specific, including about the parts a conventional credit line would quietly round off.

Human

Frank C. Langbein -- concept and brief; reviewing the material and giving advice on it; releasing, maintaining and supporting the course; inputs; the second-order reasoning material in Unit 18 and Second-Order Reasoning: The Lens Behind The Course, which derives from his own talk on the subject. Contact: frank@langbein.org.

What the release promises. The course is released and maintained by Frank C. Langbein, who answers for it in the sense this course uses: corrections and challenges go to him, and he responds. It comes with no warranty beyond what it states. The licences disclaim warranties (CC BY-SA 4.0 Section 5; AGPL-3.0 sections 15-16), and the course's own method applies to it: check what you rely on.

Getting in touch. That address is the way to reach the course: a correction, a question about a claim, an accessibility problem, or a note that something has gone out of date. The repository is published as source rather than as a place to file a report -- a reader has no account there -- so this is the route.

He has not written the course's learning material. The scripts, slides, student pages, activities, answer keys, lab code, self-check framework, animations, site, build system and narration pipeline were made by the AI systems listed below, working together; he reviewed them and gave advice. There is no exception: no unit is narrated or presented by him, and no voice in the videos is his or anyone's.

This distinction is the point rather than a caveat. A course about disclosure that blurred authorship at its own front door would be teaching one thing and demonstrating another.

Orchestration

Baba Yaga V10 -- an unreleased reasoning engine, which drove the integration: directing the AI systems below, verifying results, routing work between them, running tools, and maintaining consistency across the material.

Generating systems

Material in this repository was drafted by:

They worked through agent tools that read and edit the repository and run its commands: Claude Code (Anthropic), Codex (OpenAI) and Antigravity (Google). Claude Opus 5.5, working in Claude Code, made the narration pipeline, the avatars' rigs and the videos, and the whole-course review of October 2026. In that review GPT-6.1 Sol (through Codex) checked the code and the facts and numbers, and Gemini 4 Argon and, for the last confirmation round, Gemini 3.8 Flash (through the Antigravity CLI, agy) checked teaching, slides and page-video equivalence; both worked read-only, and Claude checked each finding before anything was changed.

Their working notes and mutual review are preserved in the git history, where most commits name the systems that made them in Co-Authored-By lines, and are distilled in CHANGELOG.md, so the process stays inspectable.

Model versions are recorded because they date the work: material generated by a given system in a given month is a fact about that moment, and the freshness policy in References and source policy depends on being able to tell when something was written.

Media production

Every unit, Unit 0 included, is narrated by three synthetic voices designed for this course, each shown as a drawn robot avatar. No voice is recorded from or cloned of any person, and each is compared with reference recordings of a real voice so that it is not close to one (docs/audio-production-handbook.md). The production systems, all run on the course's own hardware:

The voice briefs, prompts and settings are in scripts/narrator/narrator.toml, tools/ and docs/audio-production-handbook.md.

Selection preference, in order:

  1. Tools that can be downloaded and run locally.
  2. Self-hosted on local hardware (DGX Spark) where that is materially better.
  3. Remote services, only where the quality difference is considerable and the material involves nothing confidential.

Every generated visual or synthetic voice must be disclosed to learners wherever it appears. Synthetic narration must never be presented as a person speaking.

Sources not authored here

Quoted, cited, or adapted material -- Lovelace and Menabrea (1843), the secondary historical literature, published Bernoulli tables, the formal results referenced in Second-Order Reasoning: The Lens Behind The Course, and the song lyrics quoted in Unit 18 -- is attributed in References and source policy. Third-party images and media are recorded in docs/media-assets.md with source and licence.

How to cite this course

Langbein, F. C. (2026). AI Literacy: Ada, AI and Verification. Cardiff University. Made by AI systems (Claude Fable 5 / Opus 4.8, 5, 5.5, Gemini 3.1 Pro / 3.6 Flash / 3.8 Flash / 4 Argon, GPT-5.6 Sol / Terra / 6.1 Sol), orchestrated by Baba Yaga V10; reviewed, released and maintained by F. C. Langbein.

Licence

Standalone software and executable source are AGPL-3.0-or-later. Original educational content and media are CC-BY-SA-4.0. The file assignment is in REUSE.toml; the complete texts are under LICENSES/; the boundary and third-party position are in docs/licensing.md.

Third-party material -- the public-domain Lovelace and Menabrea text, and the song lyrics quoted minimally in Unit 18 under the stated fair-dealing basis -- is not covered by that licence and is attributed in References and source policy.