Joe Truss shared a great article with me, ‘The hidden cost of letting AI make your life easier‘, by Shai Tubali on Big Think. Towards the end of the article, Shai shares this:

“[Sven Nyholm’s] deeper worry is not that AI will outperform humans, but that it will appear to do so, especially to non-expert eyes. “Current forms of AI threaten meaningful activities,” he argues, “because they look far more intelligent than they are.” This appearance invites trust. People begin to treat AI as an oracle, mistaking an impressive engineering achievement for understanding. As misplaced confidence grows, judgment weakens. Skills develop less fully. Capacities are handed over too easily, and with them, forms of meaning that depend on effort.

Nyholm links this directly to the value of processes, including confusion, detours, and lingering with complexity. He punctures the idea that everything should be fast and efficient. Speed may feel pleasant, he concedes, yet it undermines patient thinking and reconsideration. He points to an Anthropic advertisement promising a paper completed in a single day: brainstorming in the morning, drafting by noon, polishing by afternoon. What disappears in this vision is the slow work of searching, getting lost, following the wrong thread, and returning with insight. “Many ideas,” Nyholm says, “come from looking for one thing and finding something else instead.” When AI delivers tidy, unified answers, it spares us that work. In doing so, it risks weakening our capacity to break complex problems into parts, examine assumptions, and think things through with precision.”

AI reduces the productive effort and struggle that makes both learning and understanding stick. Accessing information is profoundly different than understanding information, and directs the learner towards an answer instead of a learning process. This article reinforced some ideas I’ve already shared.

In ‘Keeping the friction‘ I said, “Decreasing the challenge doesn’t foster meaningful learning. Reducing the required effort doesn’t make the learning more memorable. Encouraging deeper thinking is the goal, not doing the thinking for you.”

And in ‘What’s the real AI risk in education?‘ I said, “Real learning has a charge to it, it needs to come with some challenge, and hardship. If the learning experience is too easy, it won’t be remembered. If there isn’t enough challenge, if the answers are provided rather than constructed, the learning will soon be forgotten. Remove being stuck, struggling, and failure, and you’ve removed the greatest part of a learning experience.”

I see this in my own learning. There are times I sit and read a full article, like the one shared above, but there are other times that I don’t bother and just throw a long article into an LLM and ask for a bulleted summary of the key ideas. However, I remember articles I read far better than articles where I only read the AI summaries.

How deep would my learning and understanding be if I only went as far as to read AI summaries? How much will my confidence and my belief in understanding grow, without the depth of knowledge to support my confidence and understanding? Would I be creating a kind of false fluency in topics where I lack true depth of understanding?

The convenience of using AI might not just be changing how we learn, it might be changing what we believe learning is… Perceiving learning as having access to information rather than having a deep understanding of a topic that needed to wrestle with to be truly understood. In this way, the convenience of using AI to think for us might just be reducing our intelligence.