Einstellung: The German Word That Explains Your AI Adoption Problem

By Martin Bergmann | AI Project Lab | July 2026

I was born in Germany. One word I carry into every AI adoption program I run is Einstellung, which roughly translates as mental set or attitude. It does not have a clean English equivalent, which is fitting, because the phenomenon it describes tends to go unnamed in most organizations, too.

In cognitive science, Einstellung refers to a specific failure mode: prior success with a solution blocks you from seeing a better one, even when that better solution is directly in front of you. The effect was first documented in chess. Grandmasters consistently failed to spot optimal moves because a familiar pattern had already been activated in working memory. The expert brain stops searching for the best solution the moment it recognizes a known one.

The mechanism is not stubbornness. It is efficiency. Pattern recognition is how expertise works. The same neural shortcut that makes experienced project managers fast is also what makes AI invisible to them.

The adoption problem no one names.

When senior PMs resist AI adoption, the instinct is to attribute it to politics, fear of replacement, or change fatigue. Einstellung suggests a simpler, less personal explanation: their expertise has created cognitive tunnels that AI cannot fit through. They already have a working solution for the problem. The new tool does not register as relevant because the problem already has an answer.

This has a direct design implication. Skill-building programs fail when the aperture does not exist first. Showing someone how to use AI for project intake does not land if their brain has already matched "project intake" to a known pattern and moved on.

The better starting point is a question, not a tutorial: "Here is something you have been doing the same way for five years. What would it look like if that approach were not available?" That question disrupts the pattern. It creates the opening. Then, the training has somewhere to land.

The second-order risk.

There is a less visible version of this problem worth naming. AI systems trained on your organization's historical data inherit your organization's Einstellung. The model learns your patterns and recommends within them, with algorithmic authority. If your project intake has always favored certain types of work, your AI-augmented intake will continue to favor them, invisibly. Your institutional blind spots do not disappear. They get automated.

This makes the design of the adoption program itself a high-stakes decision. The team building it likely carries the deepest Einstellung of all — the most experience, the most investment in current processes, the most pattern-matched assumptions about how work should flow.

The most useful person in that room is not the subject-matter expert. It is the person with the least prior investment in how things currently work.

Leave that person out, and your adoption program will be designed by Einstellung.