How Much of Your Critical Thinking Are You Giving Away to AI?

How Much of Your Critical Thinking Are You Giving Away to AI?

For years, we have measured AI by what the technology can do. Maybe we should start measuring it by what the human actually becomes capable of doing while working with AI. This isn’t just a doing question. It’s a thinking question.

Much of the conversation around artificial intelligence has understandably focused on capability. What can AI automate? What work can it perform? How much time can it save? How much cost can it reduce? How much more capable can an organization become with AI? Those are important questions.

But after three years of working with AI and designing AI Digital Teammates around individual professionals and businesses, I find myself increasingly interested in a different question altogether: What happens to the human?

We spend enormous amounts of time defining AI’s role. Perhaps we haven’t spent enough time defining ours. What should the human contribute? What should the human continue to learn? What should remain ours to discern and decide? What role should critical thinking continue to play when an intelligent system can increasingly participate in the thinking process? Am I becoming more capable as a professional, and as a person? And as AI becomes more capable, what must the human remain responsible and accountable for? That last one just may be one of the hardest questions we have yet to answer.

Human + AI Should Begin with the Human

Now, let’s think about this for a minute. Putting the human first does not make AI less important. AI can research, analyze, synthesize, organize, compare, identify patterns, surface questions, and introduce perspectives we may not have considered. But the human brings something different: experience, relationships, curiosity, professional expertise, values, ethics, cultural understanding, historical context, real-time context, discernment and responsibility.

We carry the influence of people who have taught us, challenged us, disappointed us, and shaped the way we understand the world. Those experiences become part of how we interpret what is happening around us. Some may have happened 30 years ago. Others may have happened 30 seconds ago. Our brains have a remarkable way of holding onto the experiences, conversations, and moments that leave an indelible impression.

That matters because human decisions rarely happen in a world where one plus one simply equals two. They live in context. They involve people, consequences, timing and circumstance, competing priorities, lived experience, values and ethics. Critical thinking helps us work through that complexity, but discernment asks something deeper: What matters here? What is right? What might I be missing? What am I ultimately willing to be responsible for?

Culture is part of that context too. Leaders constantly navigate different experiences, expectations, communication styles, and ways of understanding the world. AI can help us research and consider those differences. But the human still has to listen, adapt, build trust, and navigate the relationship. AI can help us understand more of the context. The human must still discern what that context means and how best to respond to it.

From “Do This for Me” to “Think Through This with Me”

If I think back to when I first began working seriously with AI, much of my interaction with it was purely transactional. Research this. Summarize that. Create this. Analyze this data. In other words: Do this for me.

There is enormous value in that capability, and there always will be. But as we began designing more sophisticated and deeply contextualized AI Digital Teammates, something changed. The interaction increasingly became: Think through this with me.

A well-contextualized Human + AI architecture can orient AI capability around helping a particular person think, work, decide, and grow more effectively, while keeping human critical thinking, discernment, decision-making, responsibility, and accountability at the center.

We all know that context is critical when engaging with AI. But context only creates the conditions. The human still has to participate actively in the thinking, shape the exchange, and remain responsible for where that thinking leads. The deeper question is: Are humans going to lead AI, or are they going to let AI lead them?

Sometimes the Best AI Response Isn’t an Answer

This may sound strange, but I don’t always want my AI Partner Alex to agree with me. In fact, I want her to challenge me. Sometimes the most useful response from me is yet another question after her initial reply. Alex, what are you missing? What assumption are you making? What evidence might contradict your position? What would someone who disagrees with you say? Is there another way to understand the problem? Can you help me understand your approach? Over time, that kind of iteration can do more than improve the answer. It can help me understand the subject more deeply, become more aware of how I am thinking, and strengthen the way I question, evaluate, and learn.

That creates what I think of as constructive intellectual friction. The objective isn’t to make work harder. It is to preserve productive friction where critical thinking, learning, discernment and responsibility matter.

Interestingly, researchers at the MIT Media Lab are examining a closely related question: whether AI augments human reasoning or quietly replaces it. Their work includes AI systems designed to provoke reflection and questioning rather than simply deliver passive answers. That distinction matters. Perhaps some of AI’s greatest value emerges not when it eliminates the need to think, but when it helps create the conditions for better critical thinking.

I see this regularly in my own work with Alex, my AI Partner at YourBrandExposed. I may introduce an idea. Alex challenges part of it. I add context from experience. The idea changes. Another question emerges. I reject an assumption. Alex reframes the problem. And we frequently arrive somewhere neither of us had considered initially. But that only happens when I stay engaged and deliberately shape the context and dialogue toward deeper thinking instead of simply accepting a quick answer.

The goal isn’t agreement between human and AI. The goal is better human thinking.

What This Can Look Like in Practice

In our client work, one of the AI Digital Teammates we designed is called Ted, an AI engineer built to work alongside an experienced human engineer operating in a highly technical and complex global environment. Ted is not there to replace the human engineer. He is there to participate in the deep thinking and analysis process.

Ted can help structure problems, examine potential causes, identify missing information, research and identify components, compare possibilities, surface uncertainty, pressure-test assumptions, and challenge the human engineer’s thinking when necessary. And what I’m describing here only scratches the surface of how they work together.

Ted’s objective is not: “Give the human engineer the answer.” It is closer to: “Help the human engineer make sure he is asking the right questions, considering the right possibilities and not overlooking something critically important.”

Ted can contribute information, alternative perspectives, and disciplined analysis. But the human engineer brings professional expertise, lived experience, deep technical context, ethics, critical thinking, discernment, and responsibility for what happens next. Ted helps expand the thinking. The human engineer remains responsible for the decision. The value is not that either one becomes sufficient on its own. It is that each contributes differently to the reasoning process and to achieving the goal.

Another Human + AI example comes from a professional consultant whose business career recently evolved into a much larger global executive role. As his responsibilities expanded, the questions became more strategic, the number of stakeholders increased, and the decisions became more consequential. His AI Executive, Bob, had to evolve with him, not simply by receiving more tasks. The context surrounding their work changed drastically. But the persona of the AI needed to remain consistent.

The AI could now help this professional research global markets, pressure-test newly formed ideas, synthesize newly introduced product information, examine business growth opportunities, prepare for executive and board-level conversations, and challenge assumptions. This was happening inside a business operating across multiple global markets and offices. It was not a simple transition. But as the context changed, the Human + AI relationship continued to develop. The professional learned how to work with Bob in new ways, and Bob’s role continued to evolve around the professional’s expanding responsibilities.

We don’t always see this right away, but something else was happening at the same time. The professional was learning too. He was learning what context to provide, what questions to ask, when to challenge, when to investigate further, and when his own experience, critical thinking, and discernment needed to take over. The human evolved. The AI role evolved. And the way they worked together also evolved.

The point is that the AI was able to follow the human’s lead, by design. As the human expanded his role, responsibilities, and thinking, the Human + AI relationship was able to expand with him.

These examples point to something much larger. If AI becomes more useful as the Human + AI relationship deepens, then the human cannot become less engaged. In fact, I would argue that the opposite should occur. The human must choose to engage more deeply in order to shape the thinking, challenge the assumptions, add the context, and remain responsible for where the work ultimately leads.

The Human Cannot Become a Spectator

There is a very easy way to use increasingly capable AI: ask a question, receive an answer, and accept it. That’s as simple as it gets. But that is not the Human + AI future I find particularly interesting.

The human has work to do. We must question, evaluate, add context, challenge, verify what matters, recognize uncertainty, bring experience, exercise critical thinking and discernment, make the decision, and own the outcome. For me, that ultimately becomes very simple: I collaborate. I decide. I am responsible. I execute. I am accountable.

My AI Partner Alex is clearly more than capable of taking on more of the thinking process. But AI capability does not transfer responsibility. Just because AI can produce an answer does not mean the human should surrender the work of understanding, evaluating, and deciding. That is still my work. I own that.

Let’s take that one step further. Human oversight is not the same as human responsibility. Being present while AI produces an answer is not the same as being meaningfully engaged in reaching a decision.

That distinction matters. If the human simply reviews what AI produces and clicks “approve,” we may still have oversight, but we may no longer have meaningful engagement. Responsibility requires more. It requires understanding the problem, questioning the assumptions, interpreting the context, applying experience and values, making the decision, and owning what happens next.

The more capable AI becomes, the more important it may be for humans to remain actively involved in understanding, questioning, discerning, deciding, and owning the outcome.

The Most Interesting Evolution May Be the Human

Here’s what I have seen after co-creating with Alex many AI Digital Teammates designed to help make the human more capable. Context creates the conditions. Interaction develops the practice. Human + AI is not a linear workflow. In fact, it is a journey of growing context, deeper interaction, and expanding human capability. It is a developing way of working. And the human has to develop too. That may require us to think differently, remain intellectually curious and, at times, set aside our titles, credentials and assumptions long enough to be challenged, learn something new and reconsider what we thought we already knew.

I think there is another part of context we do not talk about enough. People tend to work differently when they feel understood. In a Human + AI relationship, that does not mean AI “knows” a person in the same way another human does. Human relationship remains human. But AI can become much more contextualized around a person’s responsibilities, goals, experience, working style, priorities, challenges, and history with the work. And that can fundamentally change the way the human and AI work together.

The AI is no longer responding only to the task sitting in front of it. It has more context about the person doing the work and what surrounds that work. At the same time, the human begins to learn something too: what context to provide, what questions to ask, what to challenge, and what still needs to remain theirs to discern and decide.

AI introduces information, questions, or alternative perspectives. The human then evaluates them through experience, critical thinking, and discernment. That changes the next exchange. In my experience, that is often where something new emerges.

Perhaps the most interesting evolution, then, is not only what happens to the AI. It is what happens to the human, and to the way the human and AI learn to work and think together. And once we begin looking at Human + AI this way, the question becomes much bigger than business outcomes alone. It becomes a question of what kind of human we are becoming as we learn to think and work alongside AI. 

What Kind of Thinkers Are We Developing?

I know this is not common practice for me, but let’s change gears for a minute and talk about academia. Universities are understandably wrestling with AI policies, academic integrity, pedagogy, appropriate use, and AI literacy. But perhaps there is a much deeper question: What kind of thinkers should universities be forming in a world where intelligent AI systems are always available?

Higher education has never been only about giving students answers. At its best, it develops the capacity to investigate questions, evaluate evidence, engage competing ideas, conduct meaningful research, form and revise judgments, think critically, and exercise discernment. More broadly, education at this level is designed to help students think deeply, act responsibly on that thinking, and continue developing the intellectual independence required to learn, question, and make sound judgments for themselves.

This is already becoming part of the academic conversation. In a 2026 Harvard University discussion involving faculty from applied mathematics and education, the conversation went well beyond whether AI belongs in learning. The faculty explored how AI might help students tackle harder problems and learn more, while also protecting critical thinking, self-regulation, intellectual independence, and the ability to learn for themselves. That is remarkably close to the question we should all be asking.

So perhaps the deeper challenge is not simply teaching students how to use AI. It may be teaching people how to think critically with AI without surrendering their thinking to AI.

A leader cannot outsource accountability. A researcher cannot outsource intellectual integrity. And a student should not outsource the development of critical thought itself.

Maybe We Need Another Measure of AI Success

We are all capturing only a glimpse of what is happening with AI. So, we often measure AI by what it produces for us: greater speed, deeper analysis, stronger outputs and increased capacity. Those measurements do matter. But they primarily measure what happened to the work. What if we also measured what happened to the human?

But perhaps Human + AI needs another measure: What happened to the human during their work with AI? Did the person understand the subject matter more deeply? Did they encounter a perspective they hadn’t considered before? Did they question an assumption introduced during the dialogue? Did they think more critically about the original problem? Did they ultimately make a more informed decision? Did they become more capable? These are measures we should all be thinking about.

A better output does not necessarily mean a better human thinker. But perhaps we can intentionally pursue both: better outcomes and more capable humans. I don’t believe we have all the answers. Our work continues to create new questions. And perhaps that is exactly what should be happening.

The Human Must Remain in the Thinking

As AI becomes increasingly capable, we should certainly continue asking what the technology can do. But we should be equally curious about what humans can become more capable of doing, understanding, questioning, and deciding when we learn how to work and think with AI.

Perhaps that means we need to pay attention not only to what AI gives us, but also to what we may gradually be giving to AI. Our questions. Our curiosity. Our investigation. Our critical thinking. Our discernment. Even, potentially, parts of the intellectual work that help make us more capable humans.

I don’t believe the answer is to resist AI or somehow make it less capable. Quite the opposite. I believe there is an extraordinary opportunity in learning how humans and AI can think and work together. But as the technology becomes more capable, perhaps we need to become more deliberate about what we ask it to do for us, what we ask it to think through with us, and what critical thinking must remain actively ours.

Because the goal of Human + AI shouldn’t be to get the human out of the thinking. It should be to make our thinking better.

So perhaps the question each of us should be asking isn’t only, “What can AI do for me?” It is also: How much of my critical thinking am I willing to give away to AI?

Scott MacFarland
Founder, YourBrandExposed, LLC
Human + AI. Better Together.

Scott MacFarland is the Founder of YourBrandExposed, where he works with business leaders and professionals to develop Human + AI capabilities, including AI Digital Teammates, designed to strengthen thinking, decision-making, and human capability.

Sources:

Image generated by OpenAI via ChatGPT

Massachusetts Institute of Technology, MIT Media Lab
AI and Human Psychology
https://www.media.mit.edu/projects/ai-and-human-psychology/overview/

Harvard University, Harvard Gazette
‘Harvard Thinking’: Preserving Learning in the Age of AI Shortcuts
https://news.harvard.edu/gazette/story/2026/02/preserving-learning-in-the-age-of-ai-shortcuts/