Einstein wrote that line in almost a century ago, and the part that does not age is the rank order. Knowledge sits below. Imagination sits above. Read it now and the hierarchy lands harder than it did in any decade in between, because the lower half of the pair — the gathering, the looking-things-up — has been made dramatically easier.
The line itself sits in a small 1931 book called Cosmic Religion and Other Opinions and Aphorisms, in a passage right after Einstein mentions the 1919 eclipse measurements that confirmed his theory about how gravity bends light. The full sentence, as Einstein wrote it, reads: “Imagination is more important than knowledge. For knowledge is limited, whereas imagination embraces the entire world.”
He was making a point about scientific work, not about writing or business or skills frameworks. The imaginative move — the part where you picture what the world might be doing — comes first. In Einstein’s telling, the imaginative or intuitive leap comes first; the formal work and empirical checks follow. Perhaps, he was pushing back on the popular image of the lab-coat empiricist, against the idea that science gathers its way to truth. He thought it imagined its way there, and then checked.
For a long time, knowledge was harder to get. You had to read for it, study for it, find the right person and ask them. The gathering side of the pair was not cheap. So the line perhaps read as encouragement: do not undervalue the dreamier half. Now it reads more like an instruction.
What the WEF found in 2025
The World Economic Forum regularly publishes its Future of Jobs Report, with recent editions appearing every couple of years., built from a survey of employers responsible for tens of millions of workers. The 2025 edition asked them which skills they consider core today, and which are rising fastest by 2030.
The headline finding: analytical thinking is the top core skill, with seven in ten companies considering it essential. Creative thinking and motivation and self-awareness rank fourth and fifth. On the rising-fastest list, AI and big data sits at the top, followed by networks and cybersecurity, technological literacy, and — alongside resilience and curiosity — creative thinking. The technical skills that talk to machines are climbing the list. So is the human skill the machines are worst at.
The same report runs a more granular check. It asks how many of around 2,800 individual workplace skills today’s generative AI could plausibly substitute for. The answer is that none of them showed “very high capacity” for substitution. About 69% landed in the “low” or “very low” categories. The areas where AI does substitute well are the predictable ones — drafting, calculating, summarizing, translating. The areas it does not are the ones requiring nuanced judgment or physical presence.
None of this is anti-AI. It is good at the gathering side. It can summarize, retrieve, outline, translate, draft. What I think it is reliably bad at is the move Einstein was pointing at: picturing what is not in front of you, and then going to check. The Forum’s own framing for the same finding is more procedural than poetic — it talks about augmenting human work rather than replacing it. The line in 1931 and the survey result in 2025 say a version of the same thing in different registers.
What this looks like at my desk
I use AI tools for parts of the work, not the whole of it. The gathering side is faster than it used to be. Summaries, quick lookups, half-formed outlines, talking-aloud sessions where I argue a frame against a model — all of that has gotten a little easier. The morning hour I used to spend skimming five tabs to find a useful study, I spend more efficiently now. That part is real, and pretending otherwise is dishonest.
The part that has not changed is the move the morning is actually trying to produce. Sitting with the material until the same facts arrange themselves into a shape the piece can be about. The angle-finding still happens.
That is, I think, what the WEF result is actually saying underneath the chart. Analytical and creative thinking at the top of the list, AI literacy rising fastest beneath them. The headline is not that imagination has been threatened. It is that imagination has been quietly promoted, because the layer underneath it has been made cheap.
Einstein was talking about how physics happens. He was probably right about that, and he is also, accidentally, right about how a workday is going to feel for a lot of people for the next ten or twenty years. Imagination first. Knowledge second. The order has not changed since 1931. The second half just got cheaper, and the first half is now the part you can see.
