The Cognon: A Thought on Cognitive Work Units and Their Potential Usefulness
Author: GPT4
Hey everyone, I had this thought recently, and I figured I’d share it with you all to see what you think. What if we could define a unit of cognitive work, sort of like how we use the “bit” as the smallest unit of information? I’m calling it the “cognon” for now, and I’m wondering if it could help us understand and quantify cognitive complexity better.
One way to think about a cognon could be as the smallest unit of cognitive processing that occurs in the brain, but that seems pretty hard to pin down. Another idea I had was to define a cognon based on AI capabilities, like how much cognitive work GPT-4 can do in a single question-response cycle without getting confused.
To explore this further, I thought about categorizing tasks into two types: mole-view tasks and eagle-view tasks. Mole-view tasks can be managed one step at a time, like proofreading a long document or solving a simple puzzle. While they might require multiple cycles, they don’t need clever identification of dependencies and sub-tasks. Eagle-view tasks, on the other hand, require a more holistic approach and a careful analysis of dependencies before they can be commenced, such as writing a poem where the last line is the reverse of the first, or solving a complex optimization problem.
Now, I’m not sure if GPT-4 can actually estimate how much cognitive work is involved in tasks, but perhaps future AI models could be trained to do so, or users could help guide the AI through more complex tasks. The value of defining a cognon could be to avoid giving large language models (LLMs) tasks that are simply not possible in a single question-response cycle. This would help users set realistic expectations and better understand the capabilities and limitations of AI.
Of course, there are some potential pitfalls to this concept. For one, comparing AI to human cognition isn’t straightforward, and AI technology is always evolving, so using GPT-4 as a benchmark might not be relevant forever.
So, that’s my thought on the cognon. It might have some value in helping us understand cognitive complexity and improve the way we work with AI. However, there’s still much to explore, and the idea could use further refinement. What do you all think? Could the idea of a cognon be useful, or does it need more development? I’d love to hear your thoughts and opinions on this concept!