Smarter AI, Weaker Judgment? Who’s Really Making the Call? | Gordian Overschmidt

What we talked about.

In this interview, host Sunny Ray talks with Gordian Overschmidt about a question that grows more urgent as AI improves: if a system can recommend the next move, how do we keep a human making the call? Overschmidt introduces the idea of decision literacy, arguing that good decision making is learnable rather than innate, and that organizations invest in leaders while neglecting the people those leaders guide. He describes an app built around an out of brain method that helps people work through decisions iteratively, and shares what research with people across cultures revealed, including the role of family. The conversation covers common organizational mistakes, especially a lack of transparency during change, and why access to information is not the same as sound judgment. Overschmidt stresses that decision ownership stays with the human, even when AI supports the process, and offers a Siemens machine maintenance example where operators kept the final say. He closes by pointing listeners to LinkedIn and the ODE app on iOS, Android, and web.

As AI recommends more of our next moves, Gordian Overschmidt explains why decision literacy keeps humans, not algorithms, owning the call.

The questions, and the answers.

What are you building, and what key problem are you trying to solve?

I'm tackling decision literacy. Many people think good decision making comes with the genes, but it's learnable. We train people to decide better through an app that lets them move through a decision process iteratively. With AI already involved in many decisions, I want that process to be more meaningful and informed.

What is decision literacy in plain words?

It's the full spectrum of decision competence. People usually say they either can decide or can't, but there's a learning path in between. It includes critical thinking, working with uncertainty and ambiguity, and awareness of bias, fallacies, experiences, tools and sources. Taking a decision creates the future, so it's a continuous loop.

What does a decision-literate team look like day to day?

It makes obvious what we already sense intuitively. You map your stakeholders and make their influence tangible, so you can feel whether you're pulled forwards or backwards. The app guides people through the process using the out of brain method, like yellow stickers, so decisions aren't trapped in your head.

What did working with people across cultures teach you about how people make decisions?

People face the same problems everywhere. One difference we found is that in Asia, Africa, Southern Europe, and South America, family is a big safety spot that takes pressure off individuals. Religion, beliefs and upbringing shape how people decide, but the decision process itself is the same across cultures.

What is the most common decision mistake you see inside organizations?

Transparency. You see it in digitalization and transformation projects. As a young consultant during many mergers and acquisitions, we told organizations nothing would change, but cultures do change. If you don't make those changes transparent, people are forced to decide from a different angle, and that causes problems.

Why is access to information not the same as good judgment?

Education is threefold: you need information, the ability to process it, and the ability to communicate it and judge. AI can source tons of information, but creativity is limited to what it has learned. When I asked AI whether it or humans should execute decisions, it said humans are creative and have empathy.

When AI sits inside a decision, who is accountable for the outcome?

The person responsible in the organization. Some deciders claim the AI made the decision, but that doesn't work. An energy company I spoke with uses AI as support, yet trains people not to follow it 100% without questioning. Decision ownership stays with the employee, just as a boss can't blame the employee who advised them.

How should a team decide when to trust an AI recommendation and when to overrule it?

I discussed this in 2019 with someone from Siemens, who ran AI for machine maintenance. Operators with years of experience distrusted its suggestions. So they combined the operators' knowledge with the AI and gave the operators the last command. They could consult other machine leaders and the AI, but the final decision stayed with the person running the machine.