out.o

I spent a good amount of time last week preparing material to present to my upper management. What began as an exercise in articulating a complex idea turned into a reflection on how we think, create, and learn with tools at our disposal.

I started dumping my thoughts into a Google doc, but immediately felt constrained. It was like trying to tear open a package without scissors - certainly possible, but inefficient. My thoughts were scattered. But in a text document, they could be anything but. A document forces linearity where none exists naturally. I wanted to zoom in and out of certain concepts, easily reorganize them, and show relationships between different elements, all challenging in a text-only interface. There are no adjacent thoughts in a document. There are just sentences, lined up one after another.

Frustrated, I opened LucidChart and started with a blank LucidSpark, which is their name for a virtual whiteboard, an infinite canvas.

An infinite canvas might initially seem intimidating - it's a giant blank screen, after all. But I approach it like packing for a trip: first lay out everything I want to include before thinking about the best ways to organize them. So I started by simply putting down my thoughts across the space - drawing boxes, adding icons, and making notes - without much attention to structure. Over the next couple of days, I cleared everything and started from scratch a few times, each time getting closer to what I was in my head, which was hazy but becoming more and more clear with this exercise.

The first time I did it I laid down all the conceptual "objects" I wanted to pack. The second time I started discovering relationships between them. The third time I worked on how I could represent the objects and their relationships in a visually intuitive manner, to reduce the cognitive load of the reader. After three iterations, I cleared everything one last time and started again. This time it started to fall into place. I knew what shapes I wanted, and what they represented. I picked different colors as visual cues to represent clusters of similar things. I connected things that were associated with each other, and gave the link the same color as the shapes they were connecting, to make it easy to follow. Finally, I created distinct sections with intentional separation between parts - like chapters in a book or slides in a presentation. The fragmented thoughts had now become a cohesive narrative.

When I shared it with my product partner for feedback, he said: "I feel like I flipped through a storybook". He was kind and generous. But what mattered most to me about it was that it wasn't a generic "Looks good". It wasn't a "Nice job". It was specific. He could see the story I was trying to tell, which is exactly what I was shooting for. While something short of it would still have served the goal, it would not have brought a deep sense of personal satisfaction. But now it did. It had clicked with someone else. It's a remarkably rewarding feeling when you can transfer the abstractions and mental models in your mind to someone else's with minimal loss. It can happen through verbal communication, in written form, or even through building software, which is nothing but mental models codified in a digital tool.

These moments leave me wondering about the intimate and intertwined nature of exchange that happens between what we think of as Self and external tools. While using the canvas, my mind became indistinguishable from the tool. This experience connects directly to what philosophers Andy Clark and David Chalmers proposed in 1998 with their "extended mind" theory - the idea that our cognitive processes can and do extend beyond the brain and body into the external world through tools and environment. They introduced a concept called "cognitive artifacts", devices that serve representational functions and affect human cognitive performance. These artifacts fall into two important categories:

• Complementary artifacts, enhance our abilities during use but also allow us to learn and internalize these abilities, potentially enabling us to perform the task even without the tool. Learning mathematical reasoning through the use of pen and paper is a good example of this. Through practice, we get better at performing them in our brains.
• Competitive artifacts provide superior cognitive abilities while in use but do not necessarily lead to internalized skill enhancement and might even lead to a decline in some mental faculties. Using calculators for basic arithmetic, or using Google Maps for navigation, are prime examples. The more you use them, the dumber you become in exercising those specific skills in your brain. To be clear, competitive artifacts aren't inherently bad - they free our minds for higher-level thinking - but understanding this trade-off is important for how we choose and use our tools.

From my experience, visual storytelling tools like infinite canvases function as complementary artifacts. The more I use them, the better I can visualize complex relationships mentally. When managing multiple projects or conversations, I've developed the ability to see them as nodes and relationships in my mind, connecting dots and compressing information more effectively. There are still concepts too complex to fit in my brain alone, but my everyday visualization abilities have improved - similar to how training for a marathon makes a quick jog feel effortless.

In the age of AI when generating content is easier than ever, we face an important choice between genuine cognitive growth and mere performance enhancement. Had productivity been my goal, I could have easily "one-shotted" a presentation by dumping my scattered thoughts into an LLM and accepting a semi-meaningful output in the response. But there's something valuable - even beautiful - about having a vision and struggling to bring it to life. Even if the vision is as simple as writing a blog post, preparing a talk, or even providing feedback to someone in a way that's most effective. LLMs provide immediate gratification with "good enough" results right in the first pass, and as humans with a weakness for instant rewards, we're less motivated to persist and refine our work to its fullest potential. This suggests that LLMs can function as competitive cognitive artifacts when not used mindfully - they supercharge immediate productivity but potentially increase our dependence, leaving us feeling crippled without them.

What makes pen, paper, and infinite canvas complementary, while AI tools competitive artifacts? I believe it comes down to how learning happens. Learning happens in the physical act of drawing lines on paper or dragging shapes on a screen. It happens when we think deeply about how to best represent abstract ideas, identify the relationships between them, and clarify each step for ourselves. Most importantly, it happens in the struggle from bad to good, and good to great. That's when the attentional resources are occupied and spent. LLMs accelerate the process by helping us take leaps, from one good version to another, skipping the messy and frustrating but crucial phases of bad representations, rough drafts, and terrible organization - effectively short-circuiting the learning process itself.

There's plenty of hype about LLMs unlocking superhuman levels of productivity, and they have truth to them. But what's often missing in the conversation is the value of developing ideas before turning to AI for assistance. Create that manual draft of the story you wanna tell. Capture all the key elements you want to be included. Include your personal experiences and opinions. Make that rough sketch before you ask an image generator to one-shot it from a prompt. Spend more time in the bad, rudimentary phase, before making it good. I suspect successful people already know and practice this approach regularly. It's just not the message that sells AI services.