Delegate the Known to Better Think the Unknown

I continue to reflect on how to use generative AI in our lives and its cognitive impacts. Without academic or scientific intent. Just a personal reflection that I transcribe to share it and to help me move forward in my thinking.

I am questioning my use of AI. While I had written here that the abusive and shameless use of generative AI to write “reflective” content was unbearable to me, I let myself get caught by the machine.

I used it to write ideas on a subject close to my heart. I was seduced by the speed and apparent relevance of the generated content. I nearly fell into a form of laziness that I had been decrying just weeks before.

After producing 20 pages of content, I began to reread and refine what had come out of my exchanges with the machine. And this question of transparency came back and hit me in the face. What to do with content written this way? Should I mention the use of AI to help me generate this text? Where is the boundary of this assistance?

A question still unsolvable for me, so I continue my wanderings.

I looked at or read what others were doing and thinking. I tested and restarted my reflections with the machine by creating a writing assistant whose objective is to help with content creation:

by structuring ideas, helping make them accessible and impactful, and asking the right questions to enrich and challenge thinking. It is a thinking tool before being a simple text generator.

I got help drafting the instructions for this assistant, and some great things came out, such as:

The assistant does not lecture; it proposes, suggests, questions. It should enable writing for different audiences without ever losing the precision of reasoning.

To test it, I asked it:

I want to write a text about using AI in my daily professional life as a lawyer. I use it to generate text and content using the voice dictation feature of my devices to throw out raw and unstructured ideas (sic)

What followed was a very interesting little discussion that helped me identify some key elements of my reflection:

1. What I delegate to AI is not my thinking; it is the effort of writing what I already know.

Eureka! For weeks, I had been struggling to explain that I use AI to transfer what is in my head into a document. I would visualize the process by miming it: I take something from my forehead and transfer it to my PC screen.

I found the image evocative but not necessarily understandable for everyone.

Because in reality, when we write, we share many elements: data, information, and knowledge.

And to write this paragraph, I took 30 seconds of reflection to remember a book that explained the relationship between these 3 concepts:

Data>Information>Knowledge

Data: an elementary description;

Information: the organization of data;

Knowledge: the understanding of the meaning of information;

And in fact, when I use generative AI to generate content, I am actually asking it to transcribe a series of pieces of information and knowledge that I have in order to put a reflection in its context.

The benefit of manually typing part of this context is limited. If I have to introduce myself, writing that I am 1.85 m tall and have brown eyes is useful for the reader who doesn’t know me, but this is data that holds little interest for me. By delegating the writing of this kind of element to generative AI, I speed up the transfer process.

I don’t know if this is good or bad, but I find that in many respects, it is practical and easy to implement.

2. I don’t save time; I gain discernment: I see more quickly what truly matters.

After delegating the process of writing this data and information, I arrive more quickly at what matters, at the heart of my reflection.

To borrow an excerpt from my conversation with my AI assistant:

AI doesn't think for me. But it helps me better identify where my thinking truly begins.

Isn’t the essence of thinking to explore unknown territories? Writing pages to convey things we already know is very interesting if we want to transmit knowledge. In fact, are we still thinking when we reformulate what we already know? When I reformulate what I already know, I am no longer in discovery mode; I am in pedagogy mode.

But if we want to think and question our knowledge, we necessarily need to venture into uncharted territories of reflection. We can also, to get there, explore known concepts or areas and question them in the light of new reflections.

Consequently, the preliminary writing phase that allows us to reach these uncharted zones is, in some cases, unappealing. So, if we get help from generative AI to speed up this process in order to reach the essence of the reflection, as here where I am questioning the value of the preliminary path to a reflection, is it “cheating”? Maybe. A form of laziness? Maybe too.

But if this laziness is acknowledged. If this little “hack” allowed me to get there. According to my counter, I have already written more than 900 words here, or some 5,700 characters.

In fact, it is a shortcut that I find salutary, because it allows me to express more easily and simply things that are deeper. The effort of writing this kind of note is an exploration and a reflection that opens my mind. In this sense, I think that using generative AI for this purpose might not be such a bad thing.

It's paradoxical, but that's how it is: the more the machine helps me write, the more I can focus on what doesn't write itself so easily.