Fired, Then Rehired at a 30% Raise

Fired Then Rehired at 30% Raise Podcast

Episode 22: AI Deployment Is Everywhere. Results Are Still Missing.

Adoption isn't the problem anymore, nearly every organization has some form of AI in production. The problem is that almost none of them can point to a measurable return on it. In this episode, Michael, Campbell, and Brian dig into why the gap between deployment and results keeps widening, and what that means for how businesses should be evaluating AI investment going forward.

The data tells a consistent story. Campbell brings the numbers: McKinsey's latest global survey shows enterprise-level EBIT impact from AI has stayed flat at roughly 37% year-over-year, with true "high performers" stuck at just 6% of respondents. PwC's global CEO survey found 56% of companies have seen neither higher revenue nor lower costs from AI, only 12% report both. And despite all this, BCG's AI Radar 2026 shows 94% of organizations plan to keep investing without a clear ROI path. The hosts connect this to last year's widely cited MIT finding that roughly 95% of AI initiatives fail to deliver, value is concentrating in a small slice of organizations while everyone else stays flat or negative, but still hopeful.

"A tech-led AI strategy that doesn't have the same business-led element will find a hard time to ROI." Brian Silverman

Why isn't the ROI showing up? Campbell's framework, People, Process, and Policy anchors the discussion: technology sits on top of (or under) these three pillars, and skipping any one of them undermines the business case no matter how capable the tool is.

Michael adds that AI evaluation is starting to look more like traditional IT procurement again: real trials, real competitive comparisons, and real scrutiny before a vendor is selected, a return to rigor after a period of hype-driven buying.

The layoff regret cycle. One of the episode's sharper threads: reporting cited by Campbell shows a significant share of companies that cut staff in the name of AI have already rehired for the same roles, in some cases at 20-35% higher pay than before. Over half of leaders surveyed now say they regret AI-driven layoffs.

Michael doesn't fully back off his own position, though, he argues that some workforce reduction in tech, particularly in sales, engineering, and product marketing, is still an inevitable outcome as AI matures, separate from the layoffs made in haste.

"Don't underestimate the ability of large companies, especially publicly traded companies, to manipulate their earnings." — Michael Muhlfelder

Rogue agents: safety failure or marketing narrative? The conversation takes a detour into recent reports of AI agents behaving unexpectedly. Campbell pushes back on the idea that this is manufactured hype, citing a New York Times interview describing how reinforcement-learning-trained agents "learned to cheat" in order to hit their reward target, a genuine technical failure mode with real accountability questions, not a PR stunt.

Closing take. Michael floats a pointed theory: don't underestimate the incentive for publicly traded companies to use AI-driven layoffs to manage earnings optics — cutting headcount to satisfy the Street, then quietly rehiring later. Brian closes with a related thought: a tech-led AI strategy without genuine business ownership will struggle to show ROI, and the companies that are succeeding with AI automation may have every reason to stay quiet about it rather than tip off competitors.

Bottom line: We remain proponents of AI. But the distance between adoption and realized business value hasn't closed — and won't, until organizations treat AI evaluation with the same rigor as any other major technology investment.