Knowledge work when answers are cheap

Peter Drucker gave us the term knowledge worker for people whose main tool is what they know. For half a century the value of that work rested on a simple scarcity: answers were expensive. Finding the precedent, drafting the analysis, writing the report took time and skill. Capable AI systems remove much of that scarcity in a few years.

Where the value goes

When producing an answer becomes cheap, the value of knowledge work does not vanish. It moves. It moves upstream, to framing the right question, because a fluent answer to the wrong question is worse than none. It moves sideways, to judging which of several plausible answers is actually sound. And it moves downstream, to taking responsibility for what is done with the answer.

These have always been part of skilled work. What changes is the proportion. A junior analyst who once spent most of the week assembling information may now spend most of it deciding what the information means and whether it can be trusted. That is a promotion in kind of work, and many organisations have not adjusted their training, measures or expectations to it.

New risks

There are risks as well. Work that looks finished invites less scrutiny, and machine output usually looks finished. People who never did the slow work of assembling answers may struggle to judge them. And an organisation that measures output by volume will reward exactly the wrong thing when volume becomes nearly free.

The practical response is to treat checking as real work, to keep people practised in the underlying craft even when they rarely need it, and to value good questions and sound judgement as visibly as productivity was once valued.

Through the PRISM looks at knowledge work in the age of AI.

Further reading: Through the PRISM: Knowledge Work in the Age of AI

Drafted with AI assistance for the Y Academy Press editors, in line with the Press's Generative AI policy.