As AI gets better at doing routine and predictable work, the value of human contribution is shifting. Increasingly, what we need from people is not the ability to do the same thing faster or more consistently. Machines are getting pretty good at that. What we need from people is the ability to see what others don’t see, make unexpected connections, question assumptions, exercise judgment, and imagine entirely different approaches to problems.
Cognitive diversity is important to see different ways to approach problems.
Cognitive diversity is about differences in how people process information, focus their attention, make decisions, communicate, and solve problems. It includes both neurodivergent and neurotypical people. And it isn’t the same as cultural diversity or a personality typology. It is about how people actually think and work.
A useful way to understand cognitive diversity is through the idea of a “spiky profile.”
Rather than assuming that people are relatively even across different capabilities, imagine a profile with pronounced peaks and troughs. Someone may have an exceptional ability to focus deeply, recognize patterns, think laterally, challenge assumptions, or envision possibilities, and struggle noticeably more in another area. We all have differences. Some people’s profiles are simply spikier than others.
The problem is that workplaces tend to pay a lot of attention to the troughs. When someone doesn’t communicate the way we expect, work at the pace we expect, organize information the way we expect, or participate in meetings the way we expect, we notice. Then we often try to help that person become more like everyone else. What if the peaks are exactly what we need?
Innovation requires different types of thinking.
This is especially important for innovation. Innovation isn’t one kind of thinking. Generating a radically different idea requires something different from testing it. Challenging an idea requires something different from building it. Understanding how people will experience a new product requires something different from redesigning the system that supports it.
The most innovative team, therefore, may not be a collection of people who all think well in the same way. It may be a collection of people whose strengths are quite different from one another. That sounds good in theory. In practice, it can be difficult.
People who think differently don’t necessarily work together easily. One person wants to move now; another wants more information. One person throws out unfinished ideas to see where they go; another doesn’t want to discuss an idea until it has been thoroughly thought through. One person keeps expanding the possibilities; another immediately sees everything that could go wrong. Those differences can produce innovation. They can also drive people crazy.
This is why simply putting cognitively diverse people on a team isn’t enough. The differences themselves create friction. The goal shouldn’t be to eliminate that friction, because doing so may also eliminate some of the value of having different thinkers in the room.
The goal is to keep productive friction from becoming corrosive friction.
That requires psychological safety. People have to be able to think out loud, disagree, question assumptions, and test ideas that aren’t finished yet. It also requires a shared understanding of why colleagues may approach the same problem very differently. And it requires leaders who don’t automatically interpret a different pace, communication style, or problem-solving approach as a performance problem.
Coaching can help.
Assessment can give people a language for understanding their own patterns and those of their teammates. Coaching can then help people figure out how to use their distinctive strengths, manage predictable areas of friction, and work with people whose cognitive profiles may be quite different from their own. The unit of coaching, in this case, isn’t just the individual. It’s the team as a system.
AI adds another interesting dimension.
Most workplace systems have historically required people to adapt themselves to the way the system works. AI creates the possibility of turning some of that around. We can increasingly configure tools around the way an individual actually works.
Someone who generates exceptional ideas but struggles to organize them can use AI to provide structure. Someone who needs more processing time can use it to summarize information or prepare for a fast-moving conversation. Someone who excels at deep work but finds routine administrative tasks disproportionately taxing can use AI to reduce that cognitive burden.
In that sense, AI can be a ramp, not a mirror. It doesn’t have to make everyone work more like some imagined average employee. It can scaffold areas of difficulty so that people have more opportunity to use their distinctive strengths. That is a much more interesting vision of human-AI collaboration than simply asking which human tasks AI can replace.
Human-AI collaboration suggests a different way of thinking about neurodiversity.
Much of the conversation on neurodiversity at work has appropriately focused on accommodation: removing barriers so that neurodivergent employees can participate fully and succeed. We absolutely need to continue doing that. But accommodation isn’t the end of the conversation.
We can also ask what these differences contribute. What if some of the cognitive differences organizations have historically treated as difficulties to accommodate are also sources of capability that organizations need? What if someone who is unusually persistent, intensely focused, highly skeptical, extraordinarily attentive to detail, or inclined to make connections that other people miss has exactly the kind of cognitive spike an innovation team needs at a particular moment?
This doesn’t mean that every neurodivergent person is an innovator, that neurodivergence automatically produces particular strengths, or that one cognitive profile is better than another. That would simply replace one stereotype with another.
The value is in the combination.
The challenge for leaders is to become much more intentional about that combination: understand how people think, assemble teams with genuinely different cognitive strengths, bring different thinkers into the work when their strengths are most useful, create enough psychological safety for differences to surface, coach the friction that inevitably occurs, and use AI to scaffold people where they need support. In other words, instead of trying to round off all the spikes, we can get better at designing work around them.
The spikes may be the point.
These ideas are part of work I’ve been doing with Ken Parekh, founder of NeuroVantageAI, and Rod Hart, a systems and methodology expert with Sandbox42. We recently brought them together in our poster, Thinking Without Lines: How Neuro-Informed Coaching and Divergent Teams Are Reshaping Innovation in the Age of AI, for the 2026 Stanford Neurodiversity Summit. Our framework combines cognitive assessment, neuro-informed coaching, intentional team design, and an innovation methodology that brings different cognitive strengths into different phases of the innovation process.
We’re still developing and testing these ideas. The team example in our poster is illustrative, based on our own cognitive profiles; it is not yet validated evidence that the framework produces particular outcomes. But it gives us a useful way to examine a larger question that I think will become increasingly important as AI changes work:
How do we design organizations not merely to include people who think differently, but to benefit from the fact that they do?
