And the Problem for Enterprise AI Is … Essentially Unsolved
August 26, 2025
No AI. Just a dinobaby working the old-fashioned way.
I try not to let my blood pressure go up when I read “our system processes all your organization’s information.” Not only is this statement wildly incorrect it is probably some combination of [a] illegal, [b] too expensive, and [c] too time consuming.
Nevertheless, vendors either repeat the mantra or imply it. When I talk with representatives of these firms, over time, fewer and fewer recognize the craziness of the assertion. Apparently the reality of trying to process documents related to a legal matter, medical information, salary data, government-mandated secrecy cloaks, data on a work-from-home contractor’s laptop which contains information about payoffs in a certain country to win a contract, and similar information is not part of this Fantasyland.
I read “Immature Data Strategies Threaten Enterprise AI Plans.” The write up is a hoot. The information is presented in a way to avoid describing certain ideas as insane or impossible. Let’s take a look at a couple of examples. I will in italics offer my interpretation of what the online publication is trying to coat with sugar and stick inside a Godiva chocolate.
Here’s the first snippet:
Even as senior decision-makers hold their data strategies in high regard, enterprises face a multitude of challenges. Nearly 90% of data pros reported difficulty with scaling and complexity, and more than 4 in 5 pointed to governance and compliance issues. Organizations also grapple with access and security risks, as well as data quality, trust and skills gaps.
My interpretation: Executives (particularly leadership types) perceive their organizations as more buttoned up than they are in reality. Ask another employee, and you will probably hear something like “overall we do very well.” The fact of the matter is that leadership and satisfied employees have zero clue about what is required to address a problem. Looking too closely is not a popular way to get that promotion or to keep the Board of Directors and stakeholders happy. When you have to identify an error use a word like “governance” or “regulations.”
Here’s the second snippet:
To address the litany of obstacles, organizations are prioritizing data governance. More than half of those surveyed expect strengthened governance to significantly improve AI implementation, data quality and trust in business decisions.
My interpretation: Let’s talk about governance, not how poorly procurement is handled and the weird system problems that just persist. What is “governance”? Organizations are unsure how they continue to operate. The purpose of many organizations is — believe it or not — lost. Make money is the yardstick. Do what’s necessary to keep going. That’s why in certain organizations an employee from 30 years ago could return and go to a meeting. Why? No change. Same procedures, same thought processes, just different people. Incrementalism and momentum power the organization.
So what? Organizations are deciding to give AI a whirl or third parties are telling them to do AI. Guess what? Major change is difficult. Systems-related activities repeat the same cycle. Here’s one example: “We want to use Vendor X to create an enterprise knowledge base.” Then the time, cost, and risks are slowly explained. The project gets scaled back because there is neither time, money, employee cooperation, or totally addled attorneys to make organization spanning knowledge available to smart software.
The pitch sounds great. It has for more than 60 years. It is still a difficult deliverable, but it is much easier to market today. Data strategies are one thing; reality is anther.
Stephen E Arnold, August 26, 2025
Comments
Got something to say?
-
- Subscribe to Beyond Search
Feature archive
News archive
-
