Thinking with Machines (Part 2)
Where the Literature Led Me
In the first post of this series, I introduced the question that launched my Cognition Learning Sprint:
How do humans maintain and develop cognitive capability in AI-mediated environments?
Like many learning sprints, this one began with a collection of resources. I gathered articles, research papers, podcasts, videos, notes, and reflections. Since cognition, learning, and knowledge have been recurring interests throughout much of my life, I already had a substantial collection of material before the sprint formally began.
As I continued collecting and curating resources, a pattern started to emerge.
The relationship between technology and cognition has been debated for centuries. Socrates worried that writing would weaken memory. Ironically, we only know about it because Plato wrote it down. More recently, Nicholas Carr’s well-known article Is Google Making Us Stupid? raised concerns about the impact of search engines and digital media on attention and deep thinking.
Today, similar concerns are being expressed about Artificial Intelligence.
The term that appeared most often in my reading was cognitive offloading. The basic idea is straightforward. Humans use external tools and systems to reduce mental effort. We write things down. We use calendars and reminder systems. We rely on maps, calculators, notebooks, and increasingly, digital assistants.
In itself, cognitive offloading is neither new nor inherently bad. In fact, much of modern life depends on it. Entire fields, including knowledge management, rely on the ability to capture and share knowledge beyond what any individual can remember.
What caught my attention was not the concept itself but the direction of the conversation.
Again and again, I encountered discussions of cognitive debt, dependency, deskilling, and cognitive decline. The concern was understandable. If AI systems increasingly perform tasks that humans once performed themselves, what happens to the underlying capabilities?
These concerns are legitimate. If we stop exercising certain cognitive abilities, it is reasonable to wonder whether they may weaken over time.
As the sprint progressed, I found myself following these threads. I collected more articles. I explored the literature around cognitive offloading. I paid increasing attention to the risks.
Then, after several months, I stepped back.
One of the practices I use in Learning Sprints is periodically reviewing everything I have collected. In this case, I also used AI tools to help identify recurring concepts, themes, and patterns across my notes.
Looking at the results, I noticed something that I had not fully appreciated while immersed in the day-to-day learning process.
My attention had drifted.
The original question was still present, but much of my effort had become focused on risks, decline, and what AI might cause us to lose. The conversation around cognitive offloading had gradually become the dominant lens through which I was viewing the topic.
There is nothing wrong with that. The risks deserve careful consideration.
But they were not the question that had motivated the Learning Sprint in the first place.
Ironically, it was AI that helped me recognize the drift.
The realization prompted me to return to the original inquiry. Rather than focusing exclusively on cognitive debt and decline, I wanted to better understand the conditions under which AI might support learning, creativity, problem-solving, and other forms of cognitive development.
In other words, I wanted to explore the possibility of cognitive augmentation rather than only cognitive offloading.
The next post examines that question through a more personal lens: my own experiences using AI as a thinking partner, learning companion, and writing collaborator.