Connecting Business Intelligence to Artificial Intelligence
News Roundup - January 2026
Can a robot bake better cookies than your Grandma? Probably not, especially if it tried to make them in your kitchen.
It would probably fail before putting the cookies in the oven. Even if it found a great recipe, the best ingredients, and the perfect baking times, it wouldn’t know how to get the eggs out of my refrigerator. I always use two hands when I open it; if I don’t push with one hand and pull with the other, I’ll pull the entire appliance out of the wall. Everyone in my family knows this and does the same thing. We warn houseguests about it.
That’s tacit knowledge, and your robot can’t bake cookies for you without it. Tacit knowledge is the subjective, intuitive information known only to experienced experts. I’m seeing this theme discussed by a wide range of data writers.
Can we ever convert all tacit knowledge into data?
Yann LeCun, one of the original inventors of AI, thinks so. GK Chesterton - a voice from a century ago - takes a more circumspect approach, but they both try to jolt you out of your myopic focus on the data in front of you. Other writers, like Jessica Talisman, Bill Inmon, and Bianca Nassif approach the gap from the disciplines of organizing data. “Tacit knowledge” even shows up now in studies about AI’s effect on employment.
The articles below suggest a deeper, more thoughtful approach than most. Digesting all this takes a lot of time (which you probably don’t have), so I’ve started condensing and curating the best articles and posts I’ve read on this topic for you.
Give them a read!
Yann LeCun Has Been Right About AI for 40 Years. Now He Thinks Everyone Is Wrong, by Meghan Bobrowsky, The Wall Street Journal, Nov. 14, 2025.
LeCun warns of status quo bias. He says that language models will be replaced by “world models,” systems that learn about the world around them by taking in visual information, much like a baby animal or young child does. LLMs read text to make predictions, but a world model looks for implicit data, and that’s a whitespace right now.
Chesterton’s Fence: Understanding Past Decisions, by Jared Turner, Thoughtbot.com, July 11, 2024.
Chesterton used this simple fence metaphor to represent all past decisions you don’t know about, then showed how biases guide your reactions to them. Good data engineers take time to understand the tacit knowledge that created the fences in the first place.
ChatGPT, Business Value, and Reality - Some Critical Thinking by Bill Inmon, William’s Substack, December 8, 2025.
Bill Inmon has a new message: structured business data is more important for corporate decision-making than free-form text data. It’s an obvious point - structured business transactions result from a real decision, not a fleeting notion like a comment or rating. Structured data carries more weight. Is he right? Is that the whole story? You be the judge!
Process Knowledge Management, Part I by Jessica Talisman, Intentional Arrangement, December 3, 2025.
Talisman makes a strong case for the value of tacit knowledge. It turns an AI task from “guess the most statistically probable meaning” of this input into “reference the established definition for this context.” Her nuanced examples make this a powerful series, but the real fun is hearing how her work in process engineering inspired this focus.
The Semantic Bridge Between AI and Business Logic by Bianca Nassif and Jeff Garcia, Modern Data 101, September 2025.
I love the problem these authors address: AI tools can’t navigate structured business data as well as analysts. Why? Because so much of the knowledge behind business decisions is implicit. They suggest pushing all business logic and semantic mapping down to the database. It’s a great idea, but is the tacit knowledge behind your business decisions more or less complicated than opening my refrigerator door?
Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence, by Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen, Stanford University, August 26, 2025
Here’s a new Stanford University study (shared by Frictionless Data reader Diana Nekhorosheva) that shows how AI hurts early-career workers. Why does this happen? Most of their work is highly structured, explicitly defined, and therefore repeatable by AI. The authors predict that AI may be “less capable of replacing tacit knowledge, the idiosyncratic tips and tricks that accumulate with experience.” How should people react to this? Read the next article…
How the Internet Rewired Work—and What That Tells Us About AI’s Likely Impact, by Matt Segleman, The Wall Street Journal, November 22, 2025.
When the internet emerged thirty years ago, jobs that focused on a single, repeatable task quickly disappeared. Today, most of us book our own travel rather than going through a travel agent. People adjusted to this automation by combining (”bundling”) multiple jobs into a single role. Jobs that focus on the relationship between tasks - tacit knowledge - are the future.
To remind you of this week’s data concept, enjoy How Much More by The Go-Gos, from the Frictionless Data Spotify playlist.