In one recent survey, 74% of executives said they had more confidence in AI’s advice than in that of colleagues or friends, and 44% said they would defer to its reasoning over their own insights.
Then why do we even need managers then? Just have an AI manage my tasks, apparently it is doing 74% of the work already.
If you as a manager belong in this category, then either the AI you are using is so good it can replace you (and you should be scared), or you are so bad an AI can replace you (and you should be scared).
The third symptom is the most insidious: trusting AI’s judgement over your people’s. I’ve watched leaders run their team’s objections through a chatbot and let its verdict settle the argument.
This was the default relation I had with managers for most of my life, long before AI existed. Some people just do selective hearing, they discard the parts of reality that don’t align with their construction of reality. The article uses chatbots as an example, but it can be anything, from misreading a poll (or not knowing how to read it properly and claiming they know) to “Some other guy in another team said it was easy”.
AI is not changing bad managers. It’s just making it more obvious. Which brings me to the next quote:
So, while AI may not have created new vulnerabilities, it has certainly supercharged old ones.
I fully agree.
The article suggests a solution:
So, keep asking questions, even when the answer sounds sure. Ask what AI might be wrong about. Ask what alternatives were considered and dropped. Ask your people what they can see that the model can’t.
And I fully agree. As a programmer, I do use AI. Copilot in Github is basically everywhere, even if I don’t use it I know my superiors and my team does. The solution is not to go back in time, but to learn how to use AI properly.
Claude and GPT 5.X are specialized for coding. They suggest code and alternatives. But every time I interact with AI, I do they following:
- Do not implement anything, show me a suggestion.
- I read the suggestion, write it by hand, and I question every code line that is not obvious to me. I ask “Why is this needed?”
- I read the code, and make changes as I see fit. I then ask the AI to analyze it if necessary.
If an AI suggests there is a bug, I ask “How likely or relevant is this of happening?”. Sometimes the AI has a point, many times it does not and corrects course.
I cannot stress enough how “Asking questions” is a fundamental skill of working with AI. I am so into the Socratic method of working with it, that I genuinely cannot fathom any other way of doing it.
