Listen to this article:
If you work with data, this may be the most important article you’ll read today. This week. Maybe even this year.
It’s a bold claim, I know, but our profession needs to reorient its thinking. We are a stubborn bunch, and we’ve been banging our heads against the proverbial wall for decades. Maybe if we just package Information Management and Data Governance in a veneer of Data Stewardship or Data Literacy, leadership will “get it.” Maybe our development and business partners will start caring as much about the data as they do about the applications. A new training program. A new infographic. A new slogan. Underneath, though, little has changed.
AI is now pressing the issue. AI has piqued new interest in data. Leadership is asking questions and appearing curious, but let’s be clear:
the curiosity and interest is not about the data, it’s about how data impacts AI.
If there was a way that reliable AI could be delivered without having to deal with all this data stuff they’d do it. Perhaps we can make inroads, taking advantage of even second-hand attention, but mandating Data Stewardship or Data Literacy in the face of a fundamental lack of interest is not sustainable.
Furthermore, AI is steadily encroaching on the work of data professionals. Each new model seems to automate another task: writing code, creating mappings, generating metadata, and producing reports. Imagine Pac-Man steadily consuming one responsibility after another (or, perhaps more accurately but less well known, Langoliers). AI is taking over the work that has defined our careers. Much of the conversation is focused on workforce reduction. It’s understandable. Short-sighted organizations are using AI to cut heads.
Forward-thinking organizations recognize that AI presents an opportunity to change the nature of the work.
I frequently interact with college students, and I am inevitably asked what they should be doing now to help them to get a job when they graduate. They recognize that traditional, data-oriented entry-level roles are no longer available. Professionals whose day-to-day tasks are now being delivered by AI are asking the same question, searching for a career pivot before their role is swallowed up.
The application of AI to Information Management and Data Governance has been widely promoted. New capabilities are being delivered almost continuously. We ask: what can AI deliver? How can we get it to deliver faster? How many heads can we cut?
I’d like to consider the opposite perspective:
What will AI never do?
It’s been interesting to see how that question has changed, especially recently. The main dividing line between AI and people used to be “repetitive vs. creative” or “structured vs. unstructured.” Not any longer, though, because in each case AI has become good at both.
In our podcast, The Rock Bottom Data Feed, John Ladley and I repeatedly stress the notion that “Data is Anthropology.” It’s people. This has never been more true than right now.
People must focus on the people things.
People things have always mattered more, but we spent all our time producing artifacts. We got good at it. We got comfortable with it. We created artifacts largely for each other, and we were largely OK with it, or at least resigned to it. Besides, people things require behavior change and that’s always difficult.
A quote believed to originate in a 1979 IBM training manual was rediscovered around 2022 and has gained viral popularity recently: “A computer can never be held accountable, therefore a computer must never make a management decision.” If we’re looking for things AI will never do, this seems like a good place to start.
The dividing line between AI and people is now “producing work vs. being accountable for work.”
AI can research, develop, analyze, and recommend, but humans remain accountable. Consider the current accountability challenges with self-driving cars. Who’s accountable when an accident occurs? The question has still not been definitively resolved within the legal profession. The “driver”? The person behind the steering wheel isn’t really driving. The computer that was controlling the car? The programmers that developed the self-driving application? Corporate executives? Who?
We have at least one answer to the question, “What will AI never do?”: be organizationally accountable for the consequences of its decisions. People will always be needed for that. Now I have an answer for my students.
For years data professionals have recognized that organizations need to improve their “Data Literacy.” Nearly every Data Strategy includes “Improve Data Literacy” as an objective. And, yes, it’s important that everyone share some baseline level of understanding.
Knowledge is necessary for accountability, but knowledge itself isn’t accountability.
Some people don’t like “Data Literacy” because of the implication that someone whose skills need improving is “data illiterate.” Others have suggested “Data Fluency” or “Data Proficiency.” Call it whatever you want, the level of Data Whatever-You-Call-It has to be increased generally, but we now have a capability that is even more important to develop:
Data Accountability is how data professionals stay valuable in an AI world.
Data Accountability is your best response to AI encroachment. It is where you will add value. It is your competitive advantage. It is what makes you indispensable.
Accountability cannot be taught directly, it must be practiced, but we can equip people with the tools, information, and support they need to practice it successfully.
We can teach habits that reinforce accountability. We can understand the conditions where accountability can be developed, and where it can be destroyed. We can recognize the required culture changes. We can understand accountability behaviors and accountability enablers. We can establish structures, processes, and metrics that will improve our own decision making and maximize the probability of success in the areas for which we are accountable.
Throughout the next couple of months I will be exploring Data Accountability and especially Data Accountability Training for those concerned about AI encroaching on their jobs and who want a constructive path forward. AI doesn’t eliminate the need for data professionals.
AI eliminates the excuse for spending our careers producing artifacts that AI can now produce better and faster.
