AI Can Find Knowledge. Understanding Is Still Our Job.

Here’s one way I have seen AI being used really effectively, and it has less to do with getting AI to do the thinking for us.

Think about how our path to knowledge has changed.

In the pre-Google era, if you had a question, you went to a library, found the right books, searched through them for the relevant information, and only then could you begin the real work of understanding what you had found, connecting it with what you already knew, and forming a judgment.

Google dramatically shortened the first part. It gave us the blue links. Finding potentially relevant information became much faster, but we still had to open those links, read across them, figure out what mattered and make sense of it ourselves.

AI is collapsing that first part even further.

If a typical knowledge acquisition cycle looks like this:

Question → Relevant knowledge → Understanding → Judgment

AI can dramatically shorten Question → Relevant knowledge.

But Knowledge → Understanding → Judgment is where we should be careful about taking the shortcut.

Satya Nadella introduced the idea of cognitive coverage some time back.

As AI gets better at doing more of the work, perhaps our goal shouldn’t be to remain involved in every mechanical step.

It should be to make sure we retain coverage of the parts that matter: understanding the reasoning, questioning it where necessary, and exercising judgment over what eventually gets done.