The unsung critical skill for the future of work
In the rush to adopt generative AI in mission-driven organizations, I’ve found a lot of discussion of designing automation and developing the skill of “prompt engineering,” but no conversation about reflexivity, a skill I think is absolutely critical to writing good prompts, interpreting their output with clear eyes, and acting appropriately.
What is Reflexivity?
In simple terms, reflexivity is the ability to notice your own assumptions. It’s more than just "thinking about your thinking." It’s a conscious effort to recognize how your background, values, and the very tools you use shape the knowledge you produce.
In qualitative research in social sciences, reflexivity is what allows a researcher to say, "I am not an objective observer; my presence is part of this data". In leadership, it is what allows you to take a step back and ask, "Why am I asking this question in the first place?"
The AI Sycophancy Trap
Why is this so vital for AI? Because Large Language Models (LLMs) are, by design, sycophants. They are trained to give you what they think you want to hear.
When you ask an AI, "What are the benefits of our new peer mentoring program?" it will happily generate a list of reasons why you are right. This creates a dangerous feedback loop: your existing confirmation bias meets the AI's sycophancy, bringing us to a dangerous intersection, in which we feel false validation of our assumptions.
Practicing Reflexive Leadership
Reflexivity is a human superpower that ensures AI amplifies our mission rather than undermining it. Here is how to apply it:
Question the Options, Not Just the Execution: If you are drafting a grant, don't just ask the AI to "write a compelling case for peer mentoring." Use your reflexivity to take a step back. Ask the AI: “How can we help youth in [community] overcome [barrier or problem of interest.,” "What are the potential downsides of a peer mentoring approach for this community?" or "Give me the five strongest arguments against this program's effectiveness"
Notice the "Thinking Should Be Hard" Moments: In my book, Amplify Good Work, I share a story of a "conscientious objector" who reminded me that "sometimes, thinking should be hard". Reflexivity helps you identify when you are using AI to bypass a difficult ethical or strategic tension that actually requires your full, human attention. In other words, when you are using AI as a short cut when it should be a coach, thought partner, tutor, or set aside for now.
Identify Unusual Context: Humans are excellent at recognizing when something is "unusual or concerning," even before we can name why. AI lacks this gut-level context. Reflexivity means trusting your professional judgment to override or reconsider an AI recommendation that feels technically sound but practically off.
Moving Toward Mission-First AI
You don’t need ban AI to take an ethical approach to it: reflexivity will help you bring your own judgement along for the ride. It’s the difference between an AI that helps us work faster and one that helps us do better.
Want to build a values-aligned AI strategy for your team?
Download the Mission-First AI Starter Kit: A collection of templates and guidelines designed to help you start experimenting safely without compromising community trust.
Pre-order the Book: My forthcoming book, Amplify Good Work: Effective, Ethical AI for Mission-Driven Work, provides a comprehensive framework for navigating these tensions.
Work with Me: I offer training and consulting on AI strategy and policy through Our AI Futures Lab, helping organizations develop the reflexive leadership needed for the future of work.
I asked Gemini 3 Thinking: “I Would like to draft a blog post. Here's the idea:
Reflexivity: The Leadership Skill AI Can’t Reproduce
Deep dive into "reflexivity"—the ability to notice your own assumptions. This is borrowed from ethnogaphy. I've attached soem articles about it in Science and Technology Studies. We don't need to engage with it in depth, just want to not violate what's set out there.
We want to introduce this concept to nonprofit and government leadership and practitioners: that they should seek to be conscious of their own biases and willing to take a step back. For example, if they are writing up a grant for a new peer mentoring program, instead of asking an LLM "What are the benefits of peer mentoring?" or "How did this program help based on this data?" they could take a step back and ask about the options for addressing the problem they are seeing, their pros and cons, and so on. Further, when they get a sycophantic response back from a prompt about their work, reflexivity can help them notice that it might be triggering confirmation bias.”
and then,
”Can you create a 1:1 image without text to promote this blog? I imagine something with a mirror involved.” (I knew I was going to regret that—it gave me a kind of creepy image of a woman looking blank-faced at a hand mirror.)
”can you do somehting more abstract? Something similar what I have so far at drkarenboyd.com/blog”