Large language models have flipped a switch in how we work. Tasks that once took days now take minutes. However, that acceleration comes with a tradeoff: when machines do the processing for us, we risk losing the learning that comes from doing the work ourselves.

In essence, what LLMs are really good at is processing information. They process vast amounts of information at speeds that are difficult to comprehend and they do it endlessly, without getting tired, any time you ask them to. In a very real sense, LLMs are augmenting and even replacing the original information processors we all have access to—our brains.

Prior to November of 2022, when OpenAI launched ChatGPT, if we needed a plan or a strategy or an answer, we had to come up with it ourselves. This was often a painfully slow and inefficient process.

However, during that process, when we were researching on the internet or reading a book or asking a colleague for their perspective, we were processing that information in a way that allowed it to take root in our knowledge base. We were learning, both experientially and didactically—and later, when we had the answers we were looking for, we could recall a good deal of what we absorbed along the way.

In a sense, we were learning by being intricately involved in the process of learning—not learning by proxy, which is what’s happening when an LLM is simply presenting us with the answer.

Knowledge Retention

The difference between learning with LLMs and learning through them became clear to me when I revisited an old exercise from college: writing a business plan.

As a student, I had struggled for weeks to produce something that, in hindsight, was decidedly bad. But the struggle itself taught me how to do tasks I still find valuable today: financial modeling, competitor research, market segmentation, pricing. What I learned has stayed with me in the years that have unfolded since, which is because I had built a foundation by wrestling with the material.

Years later, when I worked on a new idea with ChatGPT, the result was extraordinary: a polished, 100-page business plan created in just 25 hours of back-and-forth. But when I returned to it weeks later, much of it felt new—as if I hadn’t fully internalized the work. (I hadn't.) I realized in that moment it was because the LLM had processed the information for me. Maybe not all of it. But lots of it. Which was problematic.

To remedy this, I had to print the business plan, reread it, highlight it, annotate it—essentially retracing the steps that embed knowledge into memory.

A Solvable Problem

This isn’t an argument against LLMs. It’s a reminder that we need to adjust how we use them. Perhaps it means slowing down and spending more time with the output until it becomes input. Perhaps it means treating the LLMs answers not as finished products but as drafts that still require our processing.

We know that innovation is always messy. It gives us speed and power, but it also asks us to be clear-eyed about what we trade away. Efficiency can’t come at the expense of comprehension.

The Real Imperative

The risk isn’t that LLMs will replace our jobs. The risk is that we’ll let them replace our thinking. This is obviously a huge mistake. Because while tools change, the need for deep, sustained cognition does not. In fact, in an era where machines can generate answers instantly, the ability to slow down, absorb, and connect ideas may be the most valuable skill we have left.

So use the models. Push them. Leverage their speed and scope. But also remember to wrestle with the work yourself. Highlight, annotate, think deeply.

Because while everything else is changing, one fact remains: you still need your brain. And keeping it sharp—retaining the ability to process, recall, and connect knowledge—is the most important task of this era.