AI Efficiency Talk Is Really a Pressure Story

Simon Carver

[calm] Welcome to the show. April 2026, and the word doing all the work in corporate language right now is “efficiency.” Meta has a memo out tied to AI and efficiency. Oracle is in the latest round of cuts. Other tech firms are trimming too. And if you’re a worker hearing that word from the inside, you’re not hearing innovation. You’re hearing: somebody has decided fewer people should carry more risk.

Lachlan Reed

[skeptical] That word -- “efficiency” -- is the bit that sticks in my throat. Because on paper it sounds neat, like tightening a few bolts in the shed. In real life, it lands like a boot on the chest. Workers hear “AI strategy,” but what they feel is fear, tighter monitoring, and this weird pressure to use tools they may not even trust. It’s a bit of a dog’s breakfast, honestly.

Jack Burns

[matter-of-fact] And the framing matters. If leadership says, “We are restructuring because markets changed,” that is one argument. If leadership says, “AI now allows us to be more efficient,” that is a very different argument. The second one implies replacement, inevitability, almost physics. It makes the decision feel objective, when in many cases it is still managerial choice.

Simon Carver

[curious] That “almost physics” line is the thing, Jack. Because once AI gets named in the memo, the whole room changes shape. People start thinking, “Was my job actually automated... or was AI just the story that made the cut easier to justify?”

Jack Burns

[reflective] Precisely. AI can change workflows. It can reduce certain categories of labor. But that does not mean every AI-linked layoff is a clean substitution of software for person. Often it is a narrative bridge. It lets management say, “This is modernization,” rather than, “This is cost control.” Those are not identical statements.

Lachlan Reed

[responds quickly] And “cost control” is at least honest. “Modernization” can hide a heap. Because if you tell staff, “Use the company AI tools, move faster, prove you’re adaptable,” you’re not just changing software. You’re changing the emotional weather at work. Every keystroke starts to feel like a test. Every hesitation looks like resistance.

Simon Carver

[softly] Yeah. And there’s a private version of this that never makes the press release. It’s not just, “Can AI help me?” It’s, “If I don’t use it, do I look replaceable? If I do use it and it goes wrong, am I blamed anyway?” That’s a rotten choice set.

Jack Burns

[skeptical] Which is why accountability has to stay with leadership. You cannot mandate tool adoption, celebrate theoretical productivity, reduce headcount, and then behave as though the consequences simply emerged from technology itself. They did not. Someone designed the policy. Someone set the target. Someone chose the threshold for acceptable risk.

Lachlan Reed

[questioning tone] Let me try to say that back, because this is where even a kangaroo could trip over it. AI might change the menu of options. But the layoff is still management picking a dish. The software didn’t sneak into the boardroom and fire Dave from accounts.

Jack Burns

[dryly] Correct. The software did not fire Dave from accounts.

Simon Carver

[half-laughs, then serious] Poor Dave. But that image actually helps. Because right now the headlines can make it sound like AI is this autonomous executive roaming the halls with a spreadsheet. And it isn’t. What it may be doing is giving cover, speed, and confidence to decisions that companies already wanted to make.

Lachlan Reed

[frustrated] And meanwhile the worker gets squeezed from both sides. Publicly, they’re told this is about productivity gains. Privately, they’re watching mates disappear, watching new dashboards roll in, and getting nudged -- or shoved -- into AI workflows. That’s not a productivity story. That’s a pressure story.

Jack Burns

[calm] Yes. And when pressure becomes the operating system, judgment deteriorates. People rush. They over-rely on outputs they should verify. They hide uncertainty because uncertainty looks expensive. That is how a tool meant to assist becomes a mechanism for organizational distortion.

Simon Carver

[pauses] So that’s the tension for today. Are we watching AI genuinely replace work? In some pockets, perhaps. But are we also watching AI become the most convenient language management has found for selling old decisions in new packaging? I think we are.

Simon Carver

[curious] And the resistance matters. Because even with all this pressure, we’re seeing claims that many enterprise workers are bypassing company AI tools. Not tweaking them -- bypassing them. And a large share of workers say they haven’t used AI at all recently. That’s a massive reality check against the hype cycle.

Lachlan Reed

[surprised] “Bypassing” is the word I can’t shake. Not ignoring. Not still learning. Bypassing. That means the tool was put in the lane and workers went around it anyway. Like a GPS that keeps steering you into a creek, so by the third trip you just chuck it on mute and follow the road signs.

Jack Burns

[matter-of-fact] And that tells you the bottleneck is not merely capability. It is reliability. A system can draft quickly, summarize quickly, even produce code quickly. But if the output requires heavy verification, introduces subtle errors, or creates compliance risk, then the apparent speed gain is partly fictional. The time has not disappeared. It has moved downstream into review.

Simon Carver

[questioning tone] So when a manager says, “This tool makes the team faster,” the sharper question is: faster at what exact stage? Drafting? Prototyping? First-pass code? Because if it then hands the next person a mess to untangle, the organization hasn’t become faster. It’s just relocated the drag.

Jack Burns

[approving][precise] Exactly. Drafting speed is not the same as system throughput. That distinction is crucial. If AI gives one employee an answer in thirty seconds, but creates two hours of checking for another employee, the claimed productivity gain may be cosmetic. In engineering, in legal review, in finance, in customer communication -- anywhere accuracy matters -- that tradeoff becomes decisive.

Lachlan Reed

[reflective] I’ve felt this myself. AI is brilliant for getting me past the blank page. If I’m staring at a doc, or trying to rough out a structure, beauty. Off it goes. But if I need something I can TRUST without babying it? Different story. Sometimes it’s like having an eager apprentice who works fast, talks confidently, and occasionally installs the brake cable where the throttle should be. Helpful right up until you ride downhill.

Simon Carver

[laughs softly] The brake-cable image is staying with me. And it gets at the emotional friction too. People aren’t just resisting because they’re stubborn or nostalgic. They’re resisting because trust is expensive to build and very easy to lose. One bad output in a sensitive workflow can poison the whole relationship.

Jack Burns

[reflective] Which is why blanket adoption mandates are usually clumsy. They assume the primary obstacle is attitude. Often it is not. Often the obstacle is that workers understand the failure modes better than executives do. The person doing the actual work can see where an error would propagate, who would have to clean it up, and what the reputational cost would be.

Lachlan Reed

[skeptical] Right, and that’s the bit the board slide never shows. It shows the shiny first draft, not the poor bugger doing QA at 7:40 p.m. trying to work out whether the AI was clever or completely off its rocker. That review bottleneck is real. You don’t remove friction by moving it to the person with the least spare time.

Simon Carver

[warmly] I think there’s also a human pride angle here. People want tools that make them better, not tools that make them feel watched, second-guessed, or interchangeable. When AI actually helps -- summarizing notes, kicking off an outline, surfacing options -- people tend to use it. When it becomes a compliance theater piece, they smell it instantly.

Jack Burns

[calm] And that returns us to the layoffs. If workers are bypassing enterprise tools, and many have not used AI recently at all, then the idea that AI has already cleanly replaced broad swaths of human labor is, at minimum, overstated. What may be expanding faster than capability is the managerial story around capability.

Lachlan Reed

[softly] Which is a nasty little twist, isn’t it? The tech may be useful, sometimes genuinely useful, while the story told around it becomes harsher than the tech itself. That’s the part I reckon workers can feel in their bones.

Simon Carver

[serious] Before we go, if this episode helped you think more clearly about what’s really happening at work, like the show, subscribe, and share it with someone trying to make sense of the AI fog right now.

Jack Burns

[measured] Because the question is no longer whether AI can produce output. It can. The question is who bears the cost when that output is unreliable, and who benefits when “efficiency” becomes a reason to cut first and explain later.

Lachlan Reed

[reflective] And maybe the question hanging over all of this is dead simple: if a company says AI made the layoff necessary, but the workers don’t trust the tool enough to use it... what exactly replaced the job?