Daily Briefing — March 31, 2026

The original pitch for AI was pretty compelling. Let the machine handle the scheduling, the formatting, the busywork that eats your morning before you even get to the real stuff. You keep the judgment calls, the creative leaps, the hard thinking that actually justifies your salary. Clean division. Everyone wins. Except that is not what happened.

What happened is that we offloaded the thinking first, because the thinking is exactly what feels hardest in the moment. Cognitive friction is uncomfortable. AI makes it disappear in seconds. So we handed over the part of the work where our actual value lives, and dressed up the result in polished language and called it a deliverable. Researchers have a word for the output now: workslop. More than 40% of workers have already run into it. You probably have too, even if you did not call it that.

The pattern gets worse at the individual level. A study of 1.5 million AI conversations found a pretty consistent loop: ask what to do, accept the answer, repeat, regret. That last step is the one people miss until it is too late. When you stop wrestling with hard problems, you get worse at wrestling with hard problems. The skill atrophies quietly, and the AI gets more confident, and somewhere in there you stopped being the one making decisions.

The division of labour was always the whole game. We just forgot to protect our side of it.

The UK government has staked serious political capital and public money on AI becoming the engine of national economic growth. Keir Starmer came in talking about unleashing AI across every sector, and billions in investment commitments followed. Sounds great on paper. The problem, as The Guardian's Aisha Down has been digging into, is that a lot of those billions are what you might generously call "phantom investments." Building projects running behind schedule, spending commitments that are vague enough to drive a lorry through, and money being shovelled toward chips that could be obsolete before they even get installed.

This matters beyond the politics. When a government over-commits to a technology narrative and the underlying reality starts showing cracks, the consequences tend to ripple outward in ways most people don't see coming until they're already living with them. Public sector tech contracts slow down. Innovation funding gets pulled or redirected. The whole "AI jobs boom" story starts looking shakier.

And then there's the bubble question. If AI spending is being propped up by hype more than genuine return on investment, a correction does not just hit the investors. It hits the companies that built hiring plans around AI growth, the teams being retrained for AI adjacent roles, and frankly anyone whose employer has been quietly justifying headcount decisions based on AI productivity promises that never quite materialised.

Monday's session told two stories at once. The Dow eked out a 0.3% gain while the S&P 500 slipped 0.3% and the Nasdaq dropped 0.7%. It is a split that reflects exactly where investor anxiety is concentrated right now: tech and chips. Micron and Sandisk led the semiconductor selloff, extending last week's losses after both the Dow and Nasdaq closed in correction territory.

The bigger headline was oil. Brent crude crossed $107 and WTI held above $103 after Trump told the Financial Times he wants the US to control Iran's oil industry "indefinitely." That's a direct signal that Middle East tensions aren't cooling, and energy markets priced it in immediately.

This week is also a critical data checkpoint. JOLTS job openings, the ADP private payrolls report, and the March jobs report are all due. But markets close Friday for Good Friday, meaning the jobs report drops into a long weekend. Traders will have less time to react, which historically amplifies volatility on Thursday.