In January 1919, two months after the armistice, W. B. Yeats wrote a poem about the feeling that things are coming apart and nobody sensible is going to stop it. It describes the conversation knowledge workers are having about AI with uncomfortable precision.
Ask a good analyst, lawyer, or software engineer what their job will look like in five years and you’ll usually get a pause, then a hedge, then a change of subject. I’ve built software for twenty-five years, and I can’t answer that question about my own work with any confidence either.
Two forecasts and a crowd
The people who do answer it tend to answer loudly. In May 2025, Anthropic’s CEO Dario Amodei told Axios that AI could wipe out half of all entry-level white-collar jobs. He put unemployment at 10 to 20 percent within one to five years (Axios). It was a projection, from the head of a company that sells the technology, and he said himself that he had no exact timeline.
Four months later, the Yale Budget Lab looked at 33 months of US employment data since ChatGPT’s release and found no discernible disruption to the broader labor market (Budget Lab). That’s a backward-looking measure of one country, and its authors say better data is needed.
Both are serious sources. Neither tells you whether to retrain, specialize, switch firms, or stay put. And between them sits a crowd of louder voices with no caveats at all: learn to prompt, learn to code, go into the trades, nothing will change.
What Yeats was watching
“The Second Coming” was written with Ireland heading into war and revolutions running across Europe. It first appeared in The Dial in November 1920 (CELT, citing A. Norman Jeffares). Its best-known couplet is this:
The best lack all conviction, while the worst Are full of passionate intensity.
The best in those lines are people whose judgment would normally carry weight, losing their nerve at the moment an order comes apart. Bertrand Russell wrote a blunter version in May 1933: “the stupid are cocksure while the intelligent are full of doubt” (Quote Investigator). Both men were describing what happens when the most careful people stop believing they can affect what comes next.
Not a milquetoast
It’s tempting to read the hedging as a character flaw. English even has a word ready for it.
Caspar Milquetoast first appeared in March 1924, in H. T. Webster’s comic strip The Timid Soul in the New York World (Wikipedia; Barry Popik). Webster named him after milk toast, toast softened in warm milk, the sort of thing you served someone with a nervous stomach. Caspar would have been timid in any decade. That’s the joke: the weakness belongs to him.
Yeats meant capable people who have gone quiet because the ground moved under them. If you’ve stopped making career decisions because every option looks like it might be the wrong one, that’s closer to his diagnosis than to Webster’s cartoon.
Why the careful freeze
Knowing more makes choosing harder. The person who has used these tools seriously has seen them do a week of work in an afternoon and has also seen them invent a source with a straight face. They’ve read both forecasts. Their honest conclusion is “I can’t tell which way this goes,” and from there it’s a short step to doing nothing.
The stakes are uneven too. Someone with fifteen years of expertise, a mortgage, and a reputation has the most to lose from a wrong move. Someone with nothing invested in the current shape of the work can bet everything on a new one and lose little. The bold career advice mostly comes from the second group.
Attention widens the gap. “Your job is gone by 2028” travels further than “parts of your job will change at different speeds, and nobody knows which parts first.” The loudest forecasts start to look like the consensus because they’re the most visible.
Then the freeze makes its own decision. A careful person who waits for clarity still ends up somewhere: their employer, the market, and the next round of restructuring choose where, on someone else’s schedule.
Why certainty feels like relief
Career dread is hard to carry because it has no edges. It’s about every skill you have, a little, all the time.
That’s why the confident prescriptions pull so hard. “Learn this tool and you’re safe” gives you an explanation, a side, and a task at once. Knowledge tends to surface complexity, while a slogan removes it, and the removal feels like relief. The worst in Yeats’s couplet are louder, and they’ve also found something that makes the dread stop, at least for themselves.
The careful person can see the complexity and won’t pretend it away, so they carry the dread privately and keep postponing the choice.
The wrong unit
Part of what makes the question unanswerable is its size. “Will my job exist in five years?” has no answer anyone can give you. A job is a bundle of tasks, judgments, and relationships, and AI is taking those apart at different speeds.
“Which parts of my work does this tool do well today, and what’s left when it does?” is a smaller question, and you can answer it yourself by watching. Figuring out what the job is happens by experiment, because the forecasts disagree and will keep disagreeing.
A career bet with a stop condition
The counterweight Yeats’s best were missing has a name: provisional conviction. Hold a position firmly enough to act on it, and stay ready to find out you were wrong. Apathy gives up the courage half of that, and the loud prescriptions give up the humility half.
In practice it means writing down three things: what you’re going to do, what result would make you stop, and the date you’ll check.
Nobody careful will say “I know where my career goes from here.” The same person can say something like “For the next three months I’ll hand the first draft of every routine report to the tool and spend the saved time on the analysis nobody asks me for. If by February the drafts still need more fixing than they save, I’ll stop treating that part of my job as the exposed part.” That sentence commits. It also says, in advance, what would prove it wrong.
The stop condition gives doubt somewhere to go besides paralysis. The dread shrinks to one decision with a date on it, and a decision that size is one a careful person can make. Run a few of those and you’ll know more about what your job has become than either forecast can tell you.
Yeats watched the center fail to hold and saw a rough beast slouching toward Bethlehem. The next few years of knowledge work will be shaped partly by whoever moves first, and right now the people moving first are mostly the ones least troubled by doubt. A careful person can claim a share of it with a bet small enough to be wrong about and a date to check it.
If someone forwarded this to you, you can subscribe at andrewlewis.ca. If you’ve made a career bet like this, or you’re stuck deciding whether to, reply and tell me what it is.