What You Outsource Will Atrophy
Writer Sahil Bloom warns that handing over general thinking to AI could leave a culture of breadth with no depth.
Curiosity Chronicle · The 2 Types of Knowledge: Real vs. SurfaceAuthor Sahil Bloom had been thinking about what most personal uses of artificial intelligence have in common. Nearly all of them, he argued, come down to outsourcing something your brain used to do. Then he offered the line that carries his whole argument: "What you outsource will atrophy."
By that, Bloom meant the difference between two kinds of knowledge. Real knowledge, in his framing, has depth. It is flexible and dynamic, and it can stretch to new problems because the subject is genuinely mastered. It comes only after long, sustained practice in one pursuit. Shallow familiarity works the other way. It is rigid and fixed, good enough to get through casual conversation or a simple test, yet it collapses when pressed.
His examples were deliberately ordinary. Hire a personal chef, Bloom said, and you will probably lose some cooking skill. Hire a money manager, and you will probably track your money less closely. Hire an analyst, and you will probably become less proficient at analysis. That trade can be fine, he argued, when it involves lower leverage tasks.
The danger, he said, is when the outsourcing spreads from specific chores to general thinking. His concern is that all real knowledge could then atrophy, leaving people defined by surface knowledge, in his words, "a species of breadth, but no depth." Social pressure makes it worse, he argued, because people feel pressure to weigh in on every subject, while saying "I don't know" is frowned upon.
That leaves a choice, in Bloom's telling. Outsource everything and become like the chauffeur in his example, able to deliver the talk flawlessly yet helpless when a listener asks something new, a future he compared to the floating humans in Wall-E. Or hand off routine areas so you can focus more intensely on the ones that matter most, and become "Max Planck, smiling in the back of the room," the Nobel laureate at the center of his example.
Questions people ask
How can I tell if I have real knowledge or just surface knowledge about something?
Try explaining it without notes, examples, or jargon, and then answering a follow-up question that changes the situation. Real knowledge is flexible, so you can apply the same idea to a new case. If you can only repeat what you heard and get stuck when the question shifts, that is a sign of surface knowledge.
Does using AI always weaken thinking, or does it depend on how you use it?
It depends on how you use it. Research on cognitive offloading finds that people remember less when they know a tool will remember for them, but tools can also deepen learning when used to test, explain, and get feedback. The risk is highest when AI supplies the final answer you copy, and lowest when you draft first and use AI to critique or quiz you.
What is a practical way to outsource some work to AI without losing depth?
Keep the hard thinking for yourself and give AI the supporting jobs. Write your own first draft, solve the problem once by hand, or state your reasoning before asking for help. Then use AI to check your logic, find counterarguments, suggest sources, or handle formatting, scheduling, and summarizing.