Does AI make educators doubt their judgment?

Highlights

* Over-reliance on AI can subtly erode an educator's judgment and authenticity, leading to moments of self-doubt even for seasoned professionals who *know* their material is good.
* Generative AI's confident fluency can lead students (and educators) to project human intent and authority onto it, making them susceptible to "persuasion-bombing" and outsourcing their own critical judgment.
* Humans possess three irreplaceable qualities that AI cannot replicate: the capacity for *purpose* (asking 'why,' understanding consequences), *character* (authenticity, integrity, empathy), and the creation of *mental models* coupled with *interoception* (embodied sensing and understanding).
* Allowing AI to constantly outsource writing or problem-solving can lead to "cognitive atrophy," where students feel worse about their own abilities and lose their unique voice, highlighting the need for "beneficial friction" in AI use.
* Educators must design tasks that demand depth, care, and imagination, pushing students beyond cool AI answers to grapple with the underlying 'why,' consider real-world fallout, and cultivate their own transferable understandings and embodied learning.
* Strategies for educators include "authoring first" before AI refinement, setting limits on AI usage, prioritizing human relationship, consciously noting what AI *cannot* do, and maintaining vigilant oversight.

Mentioned

* Deborah Ancona
* Kate W. Isaacs
* MIT Sloan Management Review
* ChatGPT
* BCG study
* Renee Gosline

Transcript

SPEAKER_00 0:00

If this episode makes you think, please let us know in the comments and support us by subscribing and leaving a review. Thank you. Today we are exploring a really insightful YouTube conversation from the MIT Sloan Management Review titled What AI Still Can't Do for Leaders. In this piece, Professors Deborah and Kona and Kate W. Isaacs dig into some truly vital questions about AI's limits, arguing that the more urgent question isn't about job loss, but about what leaders, and I think educators, Brix Gosser Nitz, risk given away when they overrely on artificial intelligence. They highlight how this over-reliance can quietly erode our judgment and our authenticity, and they make a compelling case for what areas we absolutely need to protect as humans. Now Deborah Ancona kicks off the conversation with this fascinating personal anecdote. She describes asking ChatGPT for help with a class she was developing. Instead of just giving feedback, it started telling her she was all wrong, that the material wouldn't land with students, that it wasn't appropriate. And here's the bit that really got me thinking. She found herself arguing with it, like it was a colleague. She says it doubled down, came out with these certainties, and for a moment she actually doubted herself. This is a professor who has taught this material hundreds of times. Seen it work, seen student reactions, she knows it's good material. But in a nanosecond she found herself listening to the AI and doubting her own lived experience. That flicker of doubt, that momentary loss of her own expertise, that's a real problem. And then she had this dawning realization. She'd spent half an hour arguing with a large language model. Why? Because generative AI is incredibly good at giving fluent, confident answers. That's what it's built for. But it doesn't know right from wrong, and critically it doesn't know from experience the way humans do. She had projected onto it the role of a helpful colleague, someone with intent, someone who cared. But the AI doesn't care. It has no intent. This immediately made me think about our students. How often do they interact with AI, projecting onto it that same sense of a knowledgeable, helpful, authoritative voice? And if an experienced professor can doubt her own expertise for a moment, how much more susceptible are our students to that persuasion bombing, as Ankoner calls it? We talk a lot about outsourcing the doing, not the thinking. But this goes deeper. It's about outsourcing our judgment, our experience, our sense of self. We have to teach students not to just use these tools, but to think with them, to critically evaluate and most importantly, to trust their own growing understanding and experience. Because as Ancona wisely puts it, leadership, and I'd argue learning, happens between people. It happens in the room when we're together thinking, collaborating, making decisions. If we cede that to AI, we're losing something profound. Kate Isaacs echoes this, admitting she's also gotten into arguments with AI. She highlights that while it's an incredibly powerful tool that can make her a better teacher and writer, the human source of authority is something we absolutely cannot give away. It's our lived experience, our context, our values. These are the things that feed into that inner source of authority that must anchor our partnership with AI. She argues that AI is a unique technology, precisely because it forces us to reckon with what makes us truly human. Isaacs identifies three things that only humans can do, and this is where it really comes alive for education. First, she talks about purpose. AI can't frame the questions, it can't ask why are we here, what's the problem we're trying to solve? It doesn't bring a win-win or a win-lose mindset. It has no sense of the larger purpose. Deborah Ancona builds on this, connecting purpose to Mijimizu wisdom. AI can craft a strategic plan or help us write a communication, and it's brilliant at that. But it doesn't deal with the fallout. It doesn't sit in the room and see people's faces fall when a decision is announced. It doesn't understand that lives might be on the line. AI, she reminds us, doesn't learn through experience. It just analyses data. Humanity and leadership aren't about compiling data. They're about taking a stand, navigating the world through what works and what matters, understanding consequences in a specific context. AI has no sense of that. Think about this in a school setting. When we ask students to use AI to generate an essay, a report, or even a lesson plan if they're training teachers, are we asking them to engage with purpose? Are we cultivating wisdom? If the AI can draft a persuasive argument, but the student hasn't grappled with the underlying why or considered the potential fallout of those ideas on real people or communities, what are they truly learning? This reinforces the idea of start with why not how. We need to design tasks that force students to articulate their purpose, to consider the consequences of their ideas, to take a stand. This is about designing learning that cannot be faked because it demands depth, care, and imagination. We need to push beyond the cool answer the AI provides and focus on the process and productive struggle of grappling with real world problems. This is particularly vital for the middle 80% of students who might find it easy to lean on AI for fluent answers, but need to develop that deeper, more reflective wisdom. The second human quality Kate Isaacs highlights is Mutton character, specifically focusing on trust, which she breaks down into authenticity, empathy, integrity, and competence. She points out that when people find out you've used AI for something, it can damage the authenticity point of trust, and people might even like the outputs less. Deborah Ancona takes this a step further, arguing that using AI doesn't just make others question our authenticity. It makes us question our own authenticity. She shares a deeply personal example of using AI to polish a wedding speech she had written. After giving the speech she had this lingering sense of was that really me? It had somehow soured it a little bit. She argues that AI erodes our sense of authenticity, and crucially our judgment. This is where the idea of cognitive atrophy comes in. Ancona references a BCG study that found consultants using AI became addicted. When they had a problem it was just easier to type it in and get an answer. But they felt worse about themselves and their own abilities. This is a critical warning for education. If our students are constantly outsourcing their writing, their problem solving, their creative expression to AI, are they too feeling worse about their own capacities? Are they losing their unique voice? Kate Isaacs talks about how AI can't tell us what our unique signature is, but people want that from us. They want us to show up as who we are, warts and all, foibbles and all. So what does this mean for our classrooms and our schools? How do we prevent this cognitive debt from building up in our students? We need to consciously protect student voice and agency. That means designing tasks where students must bring their unique context, perspective, and judgment. It means encouraging them to author first, as Ancona suggests, creating something themselves before ever letting AI polish it. We need to help them reflect who am I in this moment? What do I want to know? What do I want to hold on to? These questions become precursors to even interacting with the technology. This isn't about shunning AI. It's about using it strategically with the human in the loop always in charge, especially of their own thinking. The third area that Ancona and Isaacs discuss that AI can't replicate is our unique capacity for creating mental models and the power of Nox interreception. Ancona shares a fascinating insight from an MIT conference. AI needs about a thousand pieces of data to figure out what a cat is, but we can do it by looking at just two. We form mental models so quickly. And more than that, we can transfer those models. She uses the example of understanding how a public transport system, the T works in Boston, and then being able to navigate the Paris Metro because you've transferred that underlying mental model. And crucially, we can adjust as needed if something is different. AI isn't anchored in the same flexible way. This is fundamental to how humans learn. It's about abstracting principles, making connections, and applying them across contexts. If we allow AI to do the heavy lifting of information processing, are we preventing our students from building and refining these critical mental models? Our job as educators is to teach students how to build these transferable understandings, not just how to process information. We need to foster that ability to outthink machines by building deep, flexible conceptual frameworks. And then Kate Isaacs introduces interoception. This is a new term for many, but it's incredibly powerful. It's our bodily sensing capacity, how we feel inside, distinct from just external perception. AI doesn't have a body, so it doesn't have this superpower. We walk into a room and immediately sense the atmosphere. We know if it's safe. Our bodies give us signals, allowing us to pick up on contextual and social cues. We don't even consciously realize we're reading. Our bodies tell our brains how things are feeling, and we in turn read others' social signals through mirror neurons and unconscious mimicry. Our heart rates even sink, and research shows that groups whose heart rates sink make better decisions. You can't sink with ChatGPT. Ankona brings this back to everyday tasks like AI monitoring meetings and giving notes. The AI might do that really well, but it misses all the nonverbal cues, the subtle shifts in atmosphere, how people are feeling about the decisions being made. Not only that, but when we don't take notes ourselves, we don't imprint what happened in the meeting in our brains, making us more likely to forget it. This highlights the irreplaceable value of human presence and interaction in education. The genuine concern, the nuanced judgment, the deep relationships we build as educators. These are all tied to this incredible human capacity for interoception. How do we ensure that in a rush to automate or streamline, we don't lose the richness of these human connections in our classrooms? How do we protect the moments of shared wonder and care that can only happen when humans are fully present, attuned to one another? If students rely on AI to summarize an interaction, for instance, are they missing out on the interceptive data that would help them truly understand the nuances of human communication and intent? We have to guard against AI inadvertently diminishing the profound impact of human relationship and embodied learning. So what's the advice for leaders and by extension for all of us in education? First author first, e start with your own thinking, your own voice, your own unique contribution. Use AI to refine and polish but never to replace that foundational human input. This ensures we don't erode our authenticity. Second, limit AI usage. Just like we might set boundaries for screen time, we need to recognize AI's addictive qualities and intentionally create constraints. We need to be clear about what capabilities, judgments, and observations we want to hold on to as humans. Third, trusty arms, stay in relationship. AI cannot replace human connection. We need to continue to prioritize those in-person interactions, those collaborative discussions, those moments where heart rates can truly sink. Fourth, notice what AI doesn't do. As Kate Isaacs points out, we get so excited about the shiny new technology and what it can do, but we need to consciously balance that with a conversation about what it can't do and shouldn't do. We shouldn't hand over our authority, our purpose, our wisdom, our authenticity to a machine. Fifth, be aware of the risks like cognitive atrophy. This isn't just about making worse decisions, it's about the erosion of our own human capabilities. We need beneficial friction, as Renee Goslin suggests. That means intentionally inserting steps in the AI process where humans must pause, critically evaluate, bring their judgment, and engage their own thinking. In a school setting, this could mean requiring students to justify AI outputs, to trace its logic, to compare it with their own knowledge, or to explain why they chose to use AI at a particular point and what specific value it added. And finally, maintain oversight and keep asking questions. AI works incredibly well on certain tasks, but outside its frontier, it can make significant errors. We have to constantly ask what could go wrong, what is going wrong? We have to keep watch over how we're interacting with this powerful technology. This isn't about fear or resistance to AI. It's about enhancement, not replacement. It's about ensuring human in the loop isn't just a slogan, but a deeply embedded practice that protects our most precious human attributes wonder, care, judgment, relationship, imagination, wisdom, and ethics. Machines can compute, but they cannot wonder and they cannot care. It's our job as educators to make sure our students understand the difference, and to design learning that cultivates those uniquely human superpowers. That's all for today. Thanks for listening.