What I am learning as I use Gen AI
This article is written in July 2025.
If you are reading this in 2026 or later, then this article will sound a bit outdated.
——–
In 1991, I was building early versions of ERP and CRM software for my clients in India. The code base was C++ and Foxpro. We chose Foxpro because the database and the programming logic were integrated. The interface of the IDE was beautiful and it was fast to program.
But it was difficult to build out the screens. You had to program them pixel by pixel. Which made iterations harder and debugging even harder.
”Invoice number” on (30,20) from top left of screen.
”Invoice date” on (30,70) from top left of screen.
And so on.
Then, I discovered a software tool called FoxView that completely revolutionized how I designed screens. It was one of the early versions of drag and drop. I would tell what fields I want on a screen select a template, and it would build the screen and then give the code which I then included in my build.
It would animate building the screen so you could enjoy the process. It was great.
But here’s the catch: the tool’s power came with a responsibility. Without critical thinking about design principles, you’d end up with a collection of poorly designed screens. The tool amplified your design thinking – both good and bad.
Today, I see a striking parallel with Large Language Models (LLMs) and AI tools.
Just as FoxView democratized screen design, LLMs are democratizing content creation, problem-solving, and countless other tasks. But the same principle applies: the quality of your output depends entirely on the quality of your input and thinking.
The Critical Thinking Renaissance
My experience with AI has taught me something unexpected: my critical thinking skills are improving with AI usage. This might seem counterintuitive – shouldn’t AI make us lazier thinkers? The opposite is true, at least for those who use it effectively.
Here’s why: straightforward prompts to LLMs are largely useless. When you fire off a quick, vague request, you’ll find yourself iterating endlessly. Slowly, you start deviating from your original premise as you chase the AI’s suggestions rather than your own vision. Before you know it, you’re lost and ready to give up.
This frustrating cycle forced me to think more deliberately about what I want before I even approach the AI. I’ve learned to:
Define my desired outcome clearly – not just what I want, but the specific characteristics, tone, and style
Specify what I don’t want – negative examples are often as valuable as positive ones
Provide concrete examples – showing rather than just telling
Consider the context – what background information does the AI need to generate something useful?
The result? Better quality output with fewer iterations. When I invest time in upfront thinking, the LLM generates something closer to what I was envisioning – in my style, not too brash, not too formal.
The Compression of Creative Cycles
The most valuable outcome of this process isn’t just better AI output – it’s the dramatic compression of the time between critical thinking and crisp execution. Previously, having a clear vision was only the beginning of a long implementation process. Now, thoughtful preparation can lead to near-instant results.
This compression has profound implications:
For individuals: We can test more ideas, iterate faster, and explore creative territories that were previously too time-consuming to pursue.
For organizations: The bottleneck shifts from execution capacity to strategic thinking and creative vision.
For society: We’re moving toward a world where good ideas can be rapidly prototyped and tested, potentially accelerating innovation across fields.
Beyond Individual Productivity
While personal productivity gains are significant, the broader value of AI lies in how it’s reshaping our relationship with knowledge work. AI tools are becoming thinking partners – not replacements for human intelligence, but amplifiers of it.
Consider how this plays out across different domains:
Writing: AI doesn’t replace the need for clear communication skills; it amplifies them. Writers who understand structure, audience, and purpose can leverage AI to focus on higher-level creative decisions.
Problem-solving: AI can rapidly generate multiple solution approaches, but selecting the right approach still requires human judgment, domain expertise, and strategic thinking.
Learning: AI can personalize explanations and provide instant feedback, but the motivation to learn and the ability to ask good questions remain distinctly human.
The Meta-Skill of AI Collaboration
What we’re really developing is a meta-skill: the ability to collaborate effectively with AI systems. This involves:
Understanding AI capabilities and limitations – knowing when to lean on AI and when to rely on human judgment
Crafting effective prompts – developing the communication skills to guide AI toward useful outputs
Critical evaluation – assessing AI-generated content for accuracy, relevance, and alignment with goals
Iterative refinement – knowing how to build on AI outputs to achieve desired outcomes
These skills are becoming as fundamental as traditional literacy in our AI-augmented world.
The Paradox of AI-Enhanced Thinking
Here’s the paradox: AI tools are making critical thinking more important, not less. As AI handles routine cognitive tasks, the premium shifts to uniquely human capabilities – creative vision, strategic thinking, ethical reasoning, and the ability to ask the right questions.
The FoxView analogy holds: just as that screen design tool required good design thinking to produce good screens, AI requires good thinking to produce good results. The tool doesn’t replace the need for human intelligence; it amplifies it.
Looking Forward
As AI capabilities continue to expand, I expect this trend to accelerate. The individuals and organizations that thrive will be those who master the art of human-AI collaboration – who can think clearly about problems, communicate effectively with AI systems, and critically evaluate the results.
The value of AI isn’t in replacing human thinking but in creating a powerful feedback loop: better thinking leads to better AI outputs, which enables even better thinking. It’s a virtuous cycle that’s just beginning to unfold.
The question isn’t whether AI will make us better thinkers – it’s whether we’ll rise to meet the challenge of thinking better with AI.
