The AI Bubble Playbook: Hype, Judgment, and Hidden Labor

Simon Carver

[warmly] Welcome to the show. A lot of leaders right now are staring at a dashboard, a demo, maybe a few clean AI outputs, and making million-dollar decisions as if the future has already arrived. I'm Simon Carver, here with Lachlan Reed, and this episode is called, "The AI Bubble Playbook: How It Rises... and How It Ends." It's a conversation about hype, corporate behavior, and what actually happens when expectations outrun reality.

Simon Carver

And joining us is Chris J. Murphy -- CJ Murphy -- author of The Last Job You'll Ever Hate and co-founder of The Human Workforce, where these podcasts are created for our growing audience of listeners trying to make sense of this moment without getting swallowed by the noise.

Lachlan Reed

[warmly] CJ, mate, good to finally get you in the chair. We've danced around your ideas on this show a fair bit -- AI hype, weird boardroom logic, all that gear -- but getting it from the bloke who's been right in the middle of it... that's better than me trying to explain it from my shed and tripping over my own boots.

Chris J. Murphy

[calm] Good to be here. And I'll start plainly: what we're seeing with AI feels new because the tools are impressive. The pattern is not new. We've seen this behavior before -- organizations encounter a promising technology, confuse potential with capability, and then build strategies around the confusion.

Simon Carver

[curious] When you say "confuse potential with capability," I want to grab that phrase. Potential and capability sound close, but they're not the same thing at all. So what exactly is getting mixed up?

Chris J. Murphy

Potential is what a system can sometimes do under the right conditions. Capability is what it can do consistently, at scale, under pressure, with consequences attached. That's the gap. A great demo creates belief. An operating model requires reliability. And companies keep treating the demo as if it's already the operating model.

Lachlan Reed

[questioning tone] So the bubble isn't just, "AI is overhyped." It's more specific than that. It's that executives see a shiny result, then act like the whole business can run on it by Tuesday. That's the bit, yeah?

Chris J. Murphy

Exactly. The bubble is made of expectations. Not algorithms. Not models. Expectations. When leadership starts believing they have found a shortcut to performance -- faster output, lower labor cost, less friction -- they stop asking the hard question, which is: what still needs human judgment for this to work safely and well?

Simon Carver

[reflective] That's the part I find haunting, actually. Because we've all seen versions of this outside AI. A tool arrives, people don't just say, "This could help." They say, "This will solve us." And the moment a tool becomes a salvation story, clear thinking gets weird.

Chris J. Murphy

It does. Because once a technology becomes a salvation story, skepticism starts to look like disloyalty. That's when organizations become vulnerable. Not when they adopt the tool -- when they stop interrogating the assumptions surrounding the tool.

Lachlan Reed

[chuckles] Yeah, and then anyone asking a normal question gets treated like they've brought a dead possum to the quarterly planning meeting. "Hang on, who's checking the outputs?" "Who's liable when this thing gets it wrong?" Pretty basic stuff, but suddenly you're the fun police.

Simon Carver

[laughs softly] The dead possum image is regrettably effective. But let me push on the pattern piece. You're not saying AI is fake. You're saying the corporate response is familiar. What's the repeated behavior?

Chris J. Murphy

[matter-of-fact] The repeated behavior is this: organizations discover a technology that appears to compress time and cost. They narrate it as substitution rather than support. Then they make structural decisions too early. They reduce people before they understand the work. They cut process before they understand the risk. They assume efficiency before they've measured quality. We've seen this pattern before, and it ends the same way: reality re-enters the room.

Lachlan Reed

"Substitution rather than support" -- that's the one that'll stick with me. Because support sounds like a teammate. Substitution sounds like, "Beauty, we've fired the crew and handed the keys to a magic toaster."

Chris J. Murphy

[dryly] And magic toasters are usually expensive. The real story isn't whether AI matters. It does. The real story is whether leaders are mature enough to deploy it without dismantling the human systems that make performance possible in the first place.

Simon Carver

[measured] Let's stay there, because this is where the conversation usually gets fuzzy. What are companies actually getting wrong in practice? Not in the keynote version -- in the real implementation version.

Chris J. Murphy

They're treating AI as a replacement engine when, in most environments, it behaves more like an augmentation tool. That means it can accelerate drafting, pattern recognition, summarization, first-pass analysis -- but it still needs oversight, correction, and human judgment. In other words, it doesn't remove responsibility. It redistributes it.

Lachlan Reed

[responds quickly] "Redistributes it" is the killer phrase there. Because if a task still needs monitoring, validating, and cleaning up after the fact, the work hasn't vanished. It's just moved house. Different room, same mess.

Chris J. Murphy

That's right. And sometimes it's a harder kind of work. You're no longer just doing the task. You're evaluating the task, checking for failure, correcting subtle errors, and absorbing the downstream risk. Leadership often misses that because the visible output arrives quickly. The hidden labor arrives later.

Simon Carver

[curious] The visible output versus the hidden labor -- that's a sharp distinction. So the executive sees speed. The worker inherits supervision. Is that fair?

Chris J. Murphy

Yes. That's a clean way to put it. The speed is real. But speed without verification is not performance. It's just acceleration. And acceleration can move you toward a better result -- or toward a bigger mistake.

Lachlan Reed

[skeptical] Right, because "faster" isn't always "better." A trail bike rolling downhill with no brakes is moving very efficiently. Doesn't mean I'd recommend it. [laughs] And I reckon this is where companies kid themselves -- they hear efficiency and automatically hear progress.

Simon Carver

I think that's exactly the trap. Efficiency is a metric. Progress is a judgment. Those are not interchangeable. If you produce more, faster, but trust drops, quality slips, and the experienced people who used to catch mistakes are gone... what have you improved, really?

Chris J. Murphy

[reflective] That's the real question. And many companies are answering it backwards. They're chasing cost reduction before capability maturity. They see a few successful outputs, assume the system is ready for scale, and make headcount cuts or budget shifts before the tool is stable enough to carry the load. That's where the damage begins.

Lachlan Reed

"Capability maturity" -- hang on, let me try to play that back. You mean the tool might be good enough for experiments, maybe even good enough for some narrow jobs, but not mature enough to build the whole shop around. Is that close?

Chris J. Murphy

[approvingly] Very close. Maturity means consistency, governance, clear boundaries, understood failure modes, and people who know how to intervene. Without those things, what companies call transformation is often just premature restructuring.

Simon Carver

And once that restructuring happens, it gets psychologically sticky, doesn't it? Because now leaders aren't just evaluating the technology. They're defending the decision. That changes the whole conversation.

Chris J. Murphy

It does. Once jobs have been cut and budgets have been reallocated, perception management starts to replace honest assessment. Organizations become invested in the story. And when you're invested in the story, it's much harder to admit the system still depends heavily on people.

Lachlan Reed

[softly] That's the sad bit. Not just that the maths was off, but that people get treated like removable parts in a spreadsheet. Then six months later someone goes, "Huh, turns out Karen from operations was the one keeping the whole circus tent up." And Karen's already gone.

Chris J. Murphy

[calm] Exactly. Institutional knowledge is easy to undervalue and very difficult to rebuild. That's why I keep coming back to this: the primary risk here is not the technology itself. It's the decisions leaders make around the technology -- especially when those decisions are driven by hype, fear, or the promise of immediate labor reduction.

Simon Carver

[reflective] Which leaves us with an uncomfortable thought. The future of work may depend less on how smart the systems become... and more on whether the people in charge can resist mistaking a powerful tool for permission to stop thinking.

Lachlan Reed

[warmly] That's a pretty solid note to sit with, hey.