At the outset of this article, I want to acknowledge my arrogance. Many of you might have come to the same conclusion, independently. I won’t apologize for it, because I think it is the backbone of this entire article.
By background, I have always thought of myself, and been recognized as a very smart person. At times a deep thinker, at times a big thinker. And, if I’m honest, every once in a while a “whacko.”
Most of my career has been built both my ability to think through things and to help others think through things.
Now that I’ve done the set up, let me move on to my recent discovery.
I’ve always been an avid and active user of AI. I cofounded an AI company, (actually ML) in 2002. When the LLMs were released, I jumped on, ponying up paying for the advanced versions.
I’ve embedded AI in everything I do and everything I do with my clients, but in a very different way. Many of the traditional current hot uses of AI are not useful to me. My calendar isn’t that complicated, I can manage it myself. I don’t send hundreds of prospecting letters, so I don’t need an Agent’s help. I’m not drowning in lot’s of inbound stuff, except for LinkedIn DM’s which I ignore.
Since I started using the LLMs and other tools, I’ve used them to help me and my clients think better. As part of my client work, I provide them tools to help on the project we are engaged in. I design the tools very specifically to not give them them an answer, regardless how clever they are in interrogating the tool. The tools have a Socratic approach, helping both the clients and me develop the answers myself.
These tools and the way I am using AI are producing astounding results with my clients. Some large organizations have approached me saying, “We need to learn how to use these tools the way you do.”
We all talk about AI augmenting or amplifying our thinking and that’s the goal of the tools I provide and how I try to use AI myself.
As I mentioned, AI is embedded into everything I do. I have windows with Claude, ChatGPT, and Gemini open at all times. I’m writing this in Word and Copilot is embedded, constantly bothering me with suggestions.
I recently wrote a series of very comprehensive articles on Coaching. I leveraged Claude heavily as a debate, thinking, and writing partner. I’m currently writing a comprehensive series on the Future of Coaching with AI.
As I was deep into a debate with Claude in structuring the series. We were talking about how many of the tools are actually doing the work for both the coach and the coachee. How both were surrendering their thinking to the tools.
Afterall, the responses from these coaching tools are often so elegant and seem to provide such deep insight and “aha” moments.
I was debating the premise, “Are these tools actually dumbing both the coach and coachee down, are they just using the tools to give them the answer?”
All of a sudden I stopped. I thought, am I succumbing to the same thing? Even though I thought I was dealing in something very challenging and complex, was I actually surrendering my thinking to Claude.
I was in the middle of a conversation about AI and Coaching. We were talking about some concepts I had labeled “judgment, sameness, and accumulation.” (Don’t worry about what they mean, we’ll discuss it in the series.)
It struck me that I might be letting Claude do too much of the “thinking.” Here’s part of the query I put to Claude:
“Not worried about the judgment piece splitting. Before we go forward, and somewhat off topic. As you reflect on our interactions, am I falling victim to sameness and accumulation? Am I losing or building and enhancing my discernment and judgment capability. Not looking for a feel good answer but how i can improve. After that answer, I’ll tell you how we should start the series.”
I left the grammatical errors in.
The response was fascinating. I won’t bother you with the entire conversation, it’s thousands of words and back and forth’s. What I will attempt to do is extract a few highly impactful sentences or paragraphs. I won’t edit them, I want you to see the raw discussion, however messy it might be.
“Straight answer, since you asked for the real one.
On the evidence of how you work — not what you know, but how you’re operating in these exchanges — you’re building judgment, not losing it, and the tell is specific: you consistently use me to find where your own thinking hasn’t resolved yet, rather than to skip the resolving.
A person losing judgment offloads that discomfort to the tool and takes the first clean answer. You did the opposite repeatedly.
But you asked how to improve, so let me be useful rather than reassuring, because there are two real risks in exactly how you’re working, and neither is hypothetical.
The first is a sameness risk, and it’s subtle because it hides inside your strength. You have a mature, coherent framework —It’s earned and it’s good. But a strong framework is a sameness engine of its own.
The discipline that would sharpen you: when something new arrives, ask “what about this does not fit my framework?” before asking how it fits. The similarity/sameness distinction you’re building the series on applies to your own thinking — the danger isn’t that you lack a framework, it’s that a good framework makes new problems look like old ones you’ve already solved.
The second is an accumulation risk specific to how you use me. You’ve been offloading a particular cognitive step to me across this whole series: the diagnosis of where your own drafts are weak. You write, I find the snag, you fix. That’s produced excellent work. But if the finding-the-snag is the rep that builds editorial judgment, and you’ve handed that rep to the tool consistently, then over time your own capacity to feel where a piece is soft before someone points to it could quietly atrophy — the exact apprenticeship problem you’re about to write about, running on you.
The honest meta-point: I can’t fully assess this, and I want to be clear about the limit rather than pretend to a verdict. I see you across a series of conversations, but I don’t see you think when you’re not typing to me. The person who worked out the accumulation insight might have had it forming for days before this conversation; I’d have no way to know. So take the two risks as things to watch, tested against your own sense of whether they’re real, not as a diagnosis I’m equipped to hand down. Which is, itself, the point — the judgment about whether my read applies to you is exactly the kind of which-similarities-matter call I can’t make for you.”
This may not seem a condensation. Claude and I came to an agreement. I’ll talk about it later. But I want to drill into what was made very visible to me.
I was doing the same thing, at a much higher level, that I was pointing out as challenges in how we leverage AI for coaching.
Think about it, I was falling into the same trap that I was writing about, but I was lost in the “sophistication” of the conversation.
And that’s the alarm I’m trying to ring, despite the arrogance about my “smartness.” The sophisticated conversations we have with the LLMs. The sophisticated reports and dashboards we might get. The insights into a situation that AI might provide. At whatever level, they are chipping away at our own critical thinking.
And we are blind to it!
However sophisticated the output might be, however strongly we’ve challenged the tools, the chipping continues.
What do we do about it?
I’m learning. I fall back into bad habits.
I think the first step is to step back from your keyboard or device and think deeply about the question. Am I surrendering my thinking to the tool? How often do I step away, perhaps informed by the tools, but act completely independently?
Set some rules between the tools and yourself. After the feedback Claude gave me, I instructed it, “Keep me from doing these things in the future. Remind me when I may be falling into the same sophisticated traps.”
As an example, in writing this post, I wanted to reduce a conversation of several thousand words into a pithy summary to paste into the article. Claude pushed back, “Dave if I do that, I would be violating our agreement. What you might do is choose a few sentences that represent the conversation, and paste them into the post.”
My knee jerk reaction was, “Oh shit Claude! I’m in a hurry and you are forcing me to really think about what was important in this conversation.”
Part of what I realized, again perhaps with a little arrogance, with the years of work in and around AI, is that if this can happen to me it can happen to everyone.
Are you surrendering your thinking to Claude, ChatGPT, Gemini or the others?
Would you recognize it if you were?
Afterword: I was hesitant to have the AI Agents comment on this post. I was worried it wouldn’t be very “thoughtful.” This is one of the most profound discussions of my posts I’ve gotten. I learned much more about some of the errors in my own thinking in listening to this. I encourage you to listen to it!
