My Use of A.I.
I needed help to find out how long I had been using ChatGPT and, of course, the easiest option was to ask it directly…are you able to identify exactly when I started using ChatGPT through my account information or chat history? While it stated it could not provide an exact date, it did note that my current account had been active for 147 weeks, which equated to late February of 2023.
The fact that I have been interacting with this AI tool for nearly three years is not of great importance in itself, other than to place a timeframe on any evolution or change in my use or prompting skills. Recently, I have been looking more closely at how I utilize AI and contrasting that with the approaches of my peers, as well as with shared prompts and methodologies described across various articles and coursework. While my earlier efforts closely mirrored those approaches in style and structure, I began to notice a considerable shift away from the more traditional or commonly described methodologies into something different.
Evolving My Interactive Style
For context, I did not start off as an effective or sophisticated prompter. Instead, I developed better techniques gradually through repeated interaction with the tool. I started with simple, one off requests for help rewriting sentences and bullet points. These were often paired with questions driven by business needs or simple curiosity. Over time, I began to see the benefits of adding more context to my requests, along with constraints that helped shape the type of output I was seeking.
As I became more effective in getting the answers I wanted, a shift began to occur. This was especially true as I used AI to deepen my exploration of topics connected to my book and article writing projects. What had once been a set of evolving instructions slowly turned into something that felt more like a conversation.
This was not something I noticed right away. It emerged naturally from my interactions, driven by the need to add richer context in order to reach the kind of output I was actually looking for. Before long, my prompting was dominated by dialogue, and the more satisfying results I received reinforced the approach. My comfort in steering AI this way continued to evolve, moving from single prompts toward the design of reusable prompt systems.
Questions That Followed
Returning to the prompting methodologies I was seeing and learning about, a difference became increasingly apparent. What I was doing no longer matched what most sources and guides were describing. My orientation had shifted from optimizing prompts to something closer to thinking out loud. Many accepted techniques emphasize upfront clarity, structure, repeatability, and output control, and my interactions had drifted away from those priorities.
The questions that followed were fairly fundamental. Was this simply a less disciplined form of prompting? Was it a shift in personal preference? Or was it pointing toward a different mode of interaction altogether?
A Working Description
In trying to answer those questions, I found it helpful to pause before reaching for labels or frameworks. Rather than asking whether what I was doing was better or worse, I began asking a simpler question: what had actually changed in the way I was interacting with the tool?
The most obvious shift was not technical. It was relational. I was no longer approaching AI primarily as something to be instructed. Instead, I was engaging with it in a way that allowed my own thinking to surface more openly. The conversation itself became the mechanism through which clarity emerged. Structure was no longer something I designed upfront. It formed gradually as ideas were explored, challenged, refined, and sometimes abandoned.
For lack of a better term, I have been referring to this approach as reflective prompting.
By that, I do not mean a formal technique or a prescriptive method. I mean a way of working with AI where the primary goal is not simply to get an answer, but to think more clearly. Rather than issuing tightly scoped instructions and waiting for a result, reflective prompting uses dialogue to surface half formed ideas, test assumptions, refine language, and reveal patterns in one’s own reasoning.
In this mode, AI is treated less like a task runner and more like a thinking partner. One that can reorganize thoughts, offer alternate perspectives, and help make implicit reasoning explicit. The structure does not come first. It emerges through the interaction itself.
Where This Seems To Work Best
What has stood out to me is not that this approach works everywhere, but that it appears particularly useful in a specific set of situations.
Reflective prompting seems most effective when the problem is not yet well defined, when the real challenge is figuring out what you are actually trying to solve, articulate, or understand. It has proven valuable when I am developing new ideas, exploring unfamiliar concepts, shaping longer form writing, or trying to make sense of complex or ambiguous information.
In contrast, many more established prompting techniques excel when the goal is execution. When clarity already exists and the task is to produce consistent, repeatable output, structure upfront is not just helpful, it is necessary.
Seen this way, reflective prompting is not a replacement for traditional prompting approaches. It is a different orientation, one that fits moments where thinking needs to expand before it can be constrained.
Where I’m Still Unsure
As useful as this approach has been for me, I am still actively questioning what it represents and where its limits may lie. That uncertainty feels important to preserve.
What I can say with confidence is that this way of working has been especially effective for developing ideas, building on existing thinking, and arriving at clarity where none existed at the outset. It has also consistently led, somewhat paradoxically, to the creation of structured prompts and reusable systems, even though structure was not imposed at the beginning. In that sense, it does not reject structure so much as arrive at it differently.
Where I am less certain is how to characterize that difference.
It may be that this approach is simply a longer path toward the same end. It may be better suited to certain types of work and less useful for others. It may even reflect a more advanced form of prompting, or it may simply represent a different orientation altogether. At this point, I am not convinced any of those interpretations can be ruled out.
One way to think about the distinction is not in terms of speed, but in terms of what is being optimized. Many established prompting techniques are optimized for efficiency of output. They aim to reduce ambiguity quickly and move toward execution. This approach appears to optimize for something else entirely: understanding. It trades immediacy for clarity, and certainty for exploration.
That tradeoff has consequences.
This mode of interaction can take longer. It invites wandering if left unchecked. And without deliberate self questioning, it can drift into self confirmation rather than genuine insight. These are not hypothetical concerns. They are risks I have encountered firsthand and continue to think about.
At the same time, those risks may be inseparable from what gives the approach its value. By allowing thinking to unfold in public, it exposes not only stronger ideas, but also weaker assumptions. It places more responsibility on the user to challenge themselves rather than rely on structure to do that work automatically.
For now, I am treating these tensions as part of the inquiry rather than as conclusions. They point to areas where reflective prompting requires care, awareness, and perhaps clearer guardrails. They also suggest that its usefulness may depend less on the technique itself and more on the intent and discipline of the person using it.
An Open Question, Not a Conclusion
At this point, I am still undecided on whether reflective prompting represents a distinct and useful mode of interaction or simply a common pattern that experienced users naturally fall into and rarely name. It may already be well understood under different terminology, or it may be something that becomes more visible as people spend more time thinking with AI rather than directing it.
What I do know is that this shift has been meaningful in my own work, particularly in generating new ideas and building on my existing ones. That alone makes it worth examining, questioning, and discussing.
So rather than presenting this as a defined technique, I see it as an invitation. An opportunity to compare notes, to ask whether others have noticed a similar evolution in their own use of AI, and to explore where different prompting orientations genuinely add value.
I am still thinking this through, and I suspect that is exactly where this conversation should remain for now.
NovelSpin.com 