We are pouring extraordinary amounts of money, electricity, infrastructure, and ambition into making artificial intelligence more capable. But how much are we investing in the people expected to use it wisely?
That imbalance sits at the heart of a provocative new essay in Psychology Today by Cornelia C. Walther. Her concern is not simply that AI is accelerating. It is that the human abilities AI is meant to extend—attention, memory, judgment, imagination, language, and agency—can weaken when assistance quietly becomes substitution.
That distinction matters. A tool can help us reach farther while still leaving us responsible for the reaching. But when the tool routinely remembers, drafts, interprets, and reasons for us, our role can shrink from creating thought to merely approving output.
The hidden cost of convenience
Walther gathers evidence across several fronts. AI-assisted work can improve immediate performance without producing equivalent learning. Heavy reliance on a relatively small group of language models may flatten linguistic variety and reinforce dominant patterns. And behind the apparent weightlessness of every prompt sits a physical system of data centers, accelerated servers, cooling equipment, water use, semiconductor production, and energy demand.
These are different problems, but they share the same structure: what feels efficient at the level of one person and one task can create a much larger dependency when repeated across institutions and cultures.
Double literacy
The essay’s most useful idea is also its most practical: we need both human literacy and algorithmic literacy.
- Human literacy means exercising attention, emotion, aspiration, reasoning, judgment, relationships, and embodied experience.
- Algorithmic literacy means understanding what AI can do, where it fails, which incentives shape it, what happens to our data, and which decisions should remain ours.
This is a wiser frame than either reflexive rejection or passive surrender. The goal is not to avoid AI. It is to use AI in ways that preserve the capabilities that make its output worth having.
The most important AI skill may be knowing when not to outsource the struggle.
Some friction is waste. Some friction is practice. Writing the first draft, sitting with uncertainty, remembering without searching, and making a judgment before asking a model are not always inefficiencies to eliminate. Sometimes they are how a mind stays alive.
Read Walther’s complete essay at Psychology Today, then consider a deceptively simple question: Which parts of your intelligence do you want AI to amplify—and which parts do you still need to practice for yourself?
Bright, Shiny Objects! 
What do you think?