There is a pattern playing out in boardrooms that no one is talking about. It is not technical or financial. It’s psychological and it may be the single biggest reason AI is underperforming at the executive level.
When leaders are presented with AI, they don’t start by asking, “Where can this technology outperform me?” Instead, they instinctively ask, “What do I already do well that I’d rather not do anymore?” That subtle shift changes everything.
We try to automate what we are already good at. Contrary to logic, we resist using AI in the areas where it could truly outperform us. Although it sounds irrational, it happens every day.
The Hidden Pattern Behind AI Adoption
At a surface level, companies appear to be embracing AI. Chatbots are deployed. Processes are automated. Efficiency improves in pockets, but underneath that progress is a deeper behavioral pattern that quietly limits impact.
Humans automate pain, tolerate ease, fight difficulty, and protect identity. This is not theory. It is observable behavior across industries and decades of technology adoption from fax machines to modern AI systems.
Here is how it plays out:
- If something is easy and annoying → we automate it
- If something is hard and frustrating → we try to prove we can do it
- If something is hard and meaningful → we refuse to give it up
- If something defines us → we protect it at all costs
And this is exactly where AI runs into resistance.
The Air Canada Example: Automating What Humans Already Do Well
Take the example of Air Canada’s AI-powered customer service. On paper, it made perfect sense. Rebooking flights is exhausting, repetitive, and emotionally draining. Customers are frustrated, and agents are under pressure. It is a job humans hate to do. So naturally, it became a prime candidate for automation.
Technically, the system worked. It handled rebooking scenarios. It processed requests. It reduced workload. However, here is the problem: humans are actually very good at this job.
When a storm disrupts travel across Toronto and spills into the U.S., the situation becomes highly dynamic. A business traveler might consider flying to a nearby city, renting a car, calling a colleague, or adjusting meetings. A parent traveling with young children has an entirely different set of constraints. These are not just logistical problems. They are human problems.
A ticket agent can interpret nuance, ask the right follow-up questions, and adjust based on context. AI, unless deeply customized, struggles with this level of variability. That gap becomes dangerous when mistakes happen.
In one widely discussed case, Air Canada’s chatbot gave a customer incorrect information about bereavement fares. The customer acted on that advice, only to be denied reimbursement. The issue escalated to court, and the customer won.
The lesson is not that AI failed. The lesson is that we asked AI to replace humans in an area where humans are still exceptional.
The Inverse Problem: Where AI Is Strongest, Humans Resist
Now flip the scenario. Where are humans not as strong, but are deeply emotionally invested?
Areas such as diagnosis, strategy, and pattern recognition across massive datasets. Consider medicine. Doctors spend over a decade training to diagnose patients. It is intellectually rewarding, meaningful, and identity-defining work. Yet, AI can already assist, or in some cases outperform, humans in identifying patterns across large volumes of medical data.
This is not new. Even in 1998, early systems existed to help doctors reach diagnoses faster. The technology was there, but adoption was limited. Why? Simple, making the diagnosis is not just a task for doctors; it is part of their identity. Doctors do not resist AI because it lacks value. They resist it because it encroaches on the part of their work they love most. The same dynamic exists in the executive suite.
The Executive Blind Spot
Executives pride themselves on judgment.
- Evaluating people
- Setting strategy
- Making high-stakes decisions
- Reading between the lines
But research and experience suggest humans are not as good at these things as we believe. However, these are the exact areas where AI can provide the most leverage.
So, what do most organizations do?
They deploy AI in:
- Customer service
- Scheduling
- Reporting
- Administrative tasks
All these uses of AI are valuable and helpful, but none of these fundamentally change how the business thinks. Meanwhile, the highest-impact use case, AI as a decision partner at the C-level, is often ignored. As one core principle emerging from this work suggests: the CEO and the C-level executives are often the constraints in the business. If those constraints are not enhanced, the business does not truly transform.
The 2×2 That Explains Everything
To simplify this behavior, consider a 2×2 framework:
Axis 1: Easy vs. Hard for Humans
Axis 2: Emotional Response (Neutral vs Identity/Pride)
This creates four predictable behaviors:
A. Low Friction (Easy + Neutral) “I can do this, but why am I still doing it?” → Eventually automated
B. Painful (Easy + Hate) “I hate this.” → Aggressively automated
C. Ego Challenge (Hard + Neutral) “I should be able to do this.” → Humans persist
D. Identity Work (Hard + Pride) “This is my thing.” → Strong resistance to AI
The problem? The biggest opportunities for AI sit in Quadrants C and D where resistance is highest.
Why Custom AI Is the Only Real Answer
This leads to a critical realization:
You cannot solve “hard for humans, easy for AI” problems with generic AI.
Generic AI is designed to be broadly useful. It lacks:
- Your business context
- Your decision patterns
- Your leadership bias
- Your data structure
Without those, it cannot step into high-level decision-making roles effectively.
Custom AI changes that.
It allows you to:
- Train AI on your company’s data and context
- Embed leadership thinking and decision frameworks
- Weight inputs based on expertise and relevance
- Move from generic answers to strategic insight
As emphasized in implementation approaches, AI must be trained on your data, your leadership, and your context, not generic inputs.
This is the difference between AI as a tool and AI as a co-CxO.
The Real Risk: Not Using AI Where It Matters
The danger is not that AI will replace executives. Rather, the danger is that executives will refuse to use AI in the areas where it can outperform them. History is clear on this.
Industries do not get disrupted because technology exists. They get disrupted because people ignore technology where it matters most.
Executives today face the same choice.
You can:
- Use AI to save time
- Use AI to reduce workload
Or you can:
- Use AI to challenge your thinking
- Use AI to improve decisions
- Use AI to remove your blind spots
Only one of those paths creates a competitive advantage.
Final Thought
This entire conversation comes down to one uncomfortable truth: We don’t avoid AI because we don’t understand it. We avoid it because it challenges the parts of ourselves we value most.
Yet, in a world where AI continues to improve, protecting identity at the expense of performance is not a long-term strategy. The leaders who win will not be the ones who automate what they hate. They will be the ones willing to let AI challenge what they believe they are best at.