The information gap has always existed. However, against the backdrop of AI, my observation is that this gap is transforming from one of "information" to one of "cognition."
A simple LLM chat interface supercharges information retrieval. It makes gathering answers more efficient - not to mention the analytical layer LLMs add, shaping responses to fit your specific needs without requiring you to manually comprehend every piece of retrieved text. Consequently, the focus has shifted away from the "information" and toward the "question."
What should you ask? While the importance of "prompting" has been mentioned millions of times, I’ve realised that the challenge goes deeper than just word choice. I believe we are witnessing the emergence of a "Cognitive Gap." This isn't just about technical skill; it’s about how our individual mental processes - our background, logic, and perspective - dictate the very limits of what we think to ask.
"Cognition" itself is complex and varies wildly depending on the user. To put it into context, Cambridge Cognition defines it as:
"The mental action or process of acquiring knowledge and understanding through thought, experience, and the senses."
To explain why writing a great prompt is such a challenge through the lens of this definition, I’ve identified four distinct groups of users based on my own observations within the tech industry and my social circles.
Group 1: The AI Insiders
This group (which includes myself) is deeply embedded in the "AI bubble." Being inside the tech industry forces us to voluntarily digest the context of these tools. We have a stronger cognitive foundation built on:
Thought: Constantly analyzing how AI impacts products.
Experience: Heavy hands-on use or close collaboration with engineers.
Senses: A heightened sense of urgency and anxiety derived from following the lightning-fast pace of development.
Group 2: The Passionate "Solutioners"
This group is fascinated by AI’s speed and efficiency. They aren't from the tech industry and lack an engineering background, but they actively use AI tools to build products from their own perspective. They possess a "driver" or "solutioner" mindset.
However, their cognitive gap is shaped by being bombarded by "shiny" social media narratives. Their experience isn't profound; they can build things, but the mechanics remain a "black box." While some argue AI will eventually maintain all code, I believe—at least for now—human assistance and guardrails are vital. Because they lack technical "common sense," they often don't know what they don't know. Without a grasp of the fundamentals, they struggle to formally evaluate tool performance.
Group 3: The Traditionalists Under Pressure
These users work in traditional industries and face top-down pressure to embed AI into their workflows for automation. Because this pressure is external, they often lack personal appreciation for the technology.
To them, AI feels "hard" or "scary." They have zero technical exposure; even a simple answer from an LLM can feel incomprehensible if it mentions terms like "Python." To bridge their cognitive gap, we must show them that AI isn't an insurmountable barrier, embedding the learning into their daily experience until they become proficient. I’ve heard people in this group say of their corporate training: "If you learn slowly enough, you don't need to learn." They feel outpaced because the training focuses on the "what" rather than the "how."
Group 4: The Unaffected
This group is the furthest away. Their day-to-day lives and jobs are currently untouched by AI, leaving them with little experience or "sense" of the technology’s trajectory.
Conclusion: Teach a Man to Fish
There is an old Chinese proverb:
"Give a man a fish, and you feed him for a day. Teach a man to fish, and you feed him for a lifetime."
Currently, we are giving Group 2 and Group 3 a lot of "fish" - new tools and shiny interfaces - without teaching them how to fish. We need to provide the fundamentals that allow them to overcome the cognitive gap.
Ultimately, the goal is to move beyond just asking "what is the question?" and start asking: "How do you develop the cognition to know which questions are worth asking?"
A Note on Perspectives:
This post is not intended to form an academic theoretical framework or a literature review; therefore, it does not follow a formal scientific method to obtain its findings. It is purely based on my own observations within my circles and offers a personal perspective on the current landscape. While I recognise the nuances and differences among individuals within these groups, this piece looks at the situation from a high-level, abstract perspective.