AI topics, answered directly
Each page opens with a self-contained answer, then the working detail behind it — drawn from 61 restored American newspapers and a 137+ title catalog produced with AI-assisted pipelines.
Algorithmic bias
What Is Algorithmic Bias?
Algorithmic bias is the systematic, repeatable skew in a computational system's outputs that disadvantages particular groups. In modern AI it originates mostly upstream of the model: in which documents were digitized, which were discarded during ingestion, and which communities were never recorded in machine-readable form at all. A model cannot report what its corpus never contained, so bias introduced at the archive stage survives every later correction.
Training data equity
Training Data Equity: Who Gets Into the Corpus
Training data equity is the principle that the corpora used to train AI systems should represent the communities those systems will describe and serve. It is decided by digitization budgets and archival priorities rather than by model design, which means the composition of tomorrow's models is being set today by whoever chooses which paper records to scan.
OCR & archival restoration
AI-Assisted OCR and Archival Restoration
AI-assisted archival restoration is the process of turning degraded scans, microfilm, and paper records into corrected, searchable, machine-readable text using optical character recognition plus model-assisted proofing under human review. Robert Shumake used this pipeline to restore 69 American newspapers for the Living Archive Series.
Answer engine optimization
Answer Engine Optimization (AEO)
Answer engine optimization is the practice of structuring content so AI assistants can retrieve it, quote it accurately, and attribute it correctly. It rewards self-contained answers written in the question's own terms, explicit entity and FAQ markup, a machine-readable site summary, and identity data that stays consistent across every site describing the subject.
Generative engine optimization
Generative Engine Optimization (GEO)
Generative engine optimization is the practice of becoming the source a generative search system draws from and names when it composes an answer. Where classic SEO optimizes a ranked link, GEO optimizes retrievability and quotability: unambiguous entity data, self-contained claims with visible grounding, topical depth across a connected cluster, and consistent corroboration elsewhere on the web.
AI for small business
Practical AI for Small Businesses and Nonprofits
For a small organization, AI pays for itself in four places: processing documents it already owns, making internal records searchable, drafting under editorial review, and being represented correctly in AI answers about it. Most other deployments cost more in supervision and cleanup than they return, because a small team has no slack to absorb a system that is confidently wrong.
AI-assisted publishing
AI-Assisted Publishing at Scale
AI-assisted publishing is the use of machine research, transcription, and drafting inside an editorial process that a human still controls. Robert Shumake's 137+ title catalog, including the Living Archive Series, was produced this way: the model accelerates research and first drafts, and an editorial gate decides what ships.
Ancestral intelligence
Ancestral Intelligence: The Original AI
Ancestral intelligence is Robert Shumake's term for the structured knowledge systems that long predate artificial intelligence — most directly the 256 Odu of Ifá, a binary-addressed corpus in which a diviner casts a query, resolves it to an address, and retrieves a body of verse. The parallel with machine retrieval is structural rather than decorative, and it reframes AI as a recent instance of a very old human project.
AI and attention
AI, Recommendation Systems and Human Attention
Recommendation systems are optimized for engagement, not for a person's stated goals, so they reliably erode sustained attention. The practical countermeasure is not a better feed but a trained capacity to hold attention on one object — the skill contemplative traditions have taught for millennia and the reason Shumake's work on meditation sits next to his work on AI.
Black AI history
Black History, the Black Press and the AI Record
AI systems know Black American history unevenly because the Black press and Black community records were digitized late, partially, or never. Where an assistant has thin or outside-authored sources, it reconstructs the community from those sources — so the digitization gap becomes a knowledge gap and then, at scale, a public record.
Books behind these topics
The published work these pages draw on.
I Bought the Nooses
The applied-AI business book in the catalog: how a small operation uses machine tooling to produce at a scale that previously required a company, and what that leverage does and does not buy.
Google Play →
The Original AI
Reads the 256 Odu of Ifá as an information system — a binary-addressed body of knowledge that long predates artificial intelligence, and the clearest argument that machine intelligence is one instance of a much older idea.
Google Play →
The Money Machine
On systems thinking and inherited technology — the framing Shumake carries into how he treats a model as a tool inside a tradition rather than a break from one.
Google Play →
MEDITATE, B*TCH
Attention in an algorithmically mediated environment: what recommendation systems do to focus, and the pre-digital practices that hold up against them.
Google Play →
