Article / November 14, 2026
Data-Centric Design Became the Bigger Idea
My path through blockchain and AI led to a more durable construction question: how should information keep its meaning across systems and decisions?
Author
Marcus Turner
Co-founder · Workflow automation and application prototypes
About the author
Technology systems thinking and practical prototyping focused on recurring information work, source boundaries, and review responsibility.
Articles by Marcus Turner
Articles
21 articlesShowing 13 to 21
Article / November 14, 2026
Data Problems Compound Before Model Problems Appear
Data cascades show how early collection, labeling, provenance, and ownership failures can compound into downstream AI risk.
Article / November 14, 2026
Efficiency Is Not the Only AI Outcome
Taylorism offers a warning for workplace AI: improve information flow without removing judgment, learning, connection, or workers' ability to challenge the system.
Article / November 14, 2026
I Had One Game Idea. AI Gave Me Five More.
Building toys and a broken dishwasher showed me where AI can offer a useful next step and where my judgment still matters. Try one small, low risk problem if you want to see whether it helps.
From The Gradual Evolution of Artificial Intelligence | Episode 28 - Marcus Turner
Read the article
Article / November 14, 2026
Prompt Engineering Became Workflow Design
Useful AI systems depend on context, retrieval, examples, evaluation, controlled tools, and human ownership, not a collection of prompt tricks.
Article / November 14, 2026
Submittal Improvement Starts With Shared Rules
A shared data exchange can support submittal coordination, but teams must first agree on states, responsibilities, routing, exceptions, and review.
Article / November 14, 2026
The Math I Thought I Would Not Need
The mathematics I once expected software to handle became a way to ask better questions about AI, incentives, uncertainty, and connected data.
Article / November 14, 2026
What an AI Summary Cannot Tell You
A long-form hearing experiment shows how AI can help navigate a transcript while still requiring source checks, neutral questions, and human judgment.
From How I Used AI to Evaluate and Summarize a Congressional Hearing
Read the article
Article / November 14, 2026
When AI Should Explore More Than One Path
Tree of Thoughts is useful when a problem has several plausible paths, but construction teams still need evidence, evaluation criteria, and human judgment.