Definition
Earned Competence: Where Durable Confidence Comes From
Psychology settled the core of this long ago. Albert Bandura showed that the strongest source of self-efficacy is the mastery experience: you attempt something hard, you struggle, you succeed anyway, and your nervous system files the evidence. Deci and Ryan put competence beside autonomy and relatedness as one of three basic psychological needs. Confidence, in the durable sense, is not a mood or an affirmation. It is largely a record of earned wins.
That record is built in exactly the territory AI now offers to handle for us. The hard first draft. The analysis you were not sure you could do. The problem that took three failed attempts. When the tool absorbs the struggle, the output arrives but the evidence never gets filed. You produced it, yet some part of you knows you did not build it.
The research is beginning to catch up to what many people privately feel. A 2026 Nature Scientific Reports study reported that heavy, passive reliance on AI at work was associated with lower self-efficacy, ownership, and experienced meaning, while active, effortful collaboration with the tool was not. Fortune has described what one Michigan researcher calls competence vertigo: accomplished professionals whose imposter feelings may, for the first time, have some factual basis, because the performance genuinely is not all theirs.
I want to be careful here, because the mental health conversation deserves precision. I will not claim AI causes the anxiety we see in young people; that story has many authors, and researchers like Jonathan Haidt have documented other forces at work. But earned competence and mental health belong in the same conversation. A generation that performs impressively while privately doubting whether any of it is theirs is not a confident generation, whatever its output suggests.
The answer is not less capability. It is protecting the earning. Keep some struggles. Let the reps that build a person stay expensive. Use the tool where output is the point, and hold the line where development is the point. Knowing which is which may be the central judgment call of this era, for parents, for managers, and for anyone building themselves.
How to protect the earning
Use AI after the first attempt, not before it. Keep some reps unaided. Make the person explain the decision path, not just hand over the result. Separate output tasks from development tasks, and defend the second kind.
Research behind this
Common questions
- Does using AI lower confidence?
- Research suggests passive reliance is associated with lower self-efficacy and ownership, while active collaboration largely is not. The difference is whether you still do the effortful part.
- Why do capable people feel like frauds in the AI era?
- Because AI can create a real gap between what a person can produce and what they feel they have actually mastered. Sometimes the feeling is not irrational insecurity. It is a signal that performance has outrun earned competence.
- How do you build earned competence deliberately?
- Genuine stakes, accountability for real decisions, experience deep enough to recognize in others, and values held when holding them costs something.