For the third week running, the gap between AI's cheerleaders and its critics holds steady. But underneath the stability, a jobs story is hardening into something that can't be spun away.
New to Fizzle, Spread, Domination, Autonomy and Chaos? Here’s why I renamed the five worlds – the Aaronson & Barak originals appear in brackets.
This week's big news is simple: more companies than ever are saying AI is the reason they're letting people go. Oracle cut 21,000 jobs this year alone. At the same time, most companies still can't point to big gains from their AI spending – only about one in four say they got what they expected. So right now we're caught between two very different futures. One is the Spread (Futurama) future, where AI is just a helpful tool that makes work better and humans stay in charge. The other is more like the Domination (AI-Dystopia) future, where AI quietly takes power away from ordinary people – not by robots taking over, but by bosses using AI to cut costs while workers pay the price. The score this week says the Spread story is still winning – but the gap is shrinking where it matters most.
The Domination–Spread gap is the single number I track hardest each week. It measures how much the optimistic, human-in-control narrative is outrunning – or being outrun by – the concentrated-power, workers-left-behind narrative.
This week the gap sits at –20 percentage points: Spread at 44%, Domination at 24%. That is almost identical to last week's –22. On the surface, nothing moved. But the series arc tells a more important story.
Cast your eye back to Scan 1 in April, when Domination led Spread by +5 points. By Scan 2 the gap had blown out to +24 in Domination's favour – the widest panic reading in the series. Then something shifted. From Scan 5 onwards, Spread crossed into the lead and has held it for seven consecutive scans. The gap peaked in Spread's favour at –22 last week. This week's –20 is the first fractional narrowing in that run. One data point doesn't make a trend. But when the underlying stories are 21,000 Oracle layoffs and AI named as the number-one reason for cuts three months running, the narrowing deserves watching.
The most significant shift in this week's scan is not a number – it's a narrative hardening. The labour displacement story has moved from prediction to documentation.
TechCrunch's running layoff tracker, Forbes on Oracle, CNBC's report that AI is now the leading stated reason for job cuts, Tom's Hardware counting nearly 80,000 Q1 losses with half attributed to AI: these are not opinion pieces. They are receipts. For the first time in this series, Domination's signal is coming primarily from employment data rather than from power-concentration abstractions or regulatory alarm.
Harvard Business Review adds a crucial wrinkle: it argues companies are cutting based on AI's potential, not its performance. That makes the displacement worse, not better. Workers are losing jobs to a technology that hasn't yet delivered what its sellers promised. That is a Fizzle (AI-Fizzle) story wearing a Domination mask – and it connects directly to the Fizzle cluster this week.
At 17%, Fizzle is broadly flat – it ranged from 7% to 19% across the series, so this reading is mid-range, not extreme. But the sources underneath it are sharper than usual. WRITER's enterprise survey found only 29% of companies see significant ROI. Terminal X puts the project abandonment rate at 42%. Larridin's developer benchmarks show real-world AI tools delivering 1.6x returns against 10x vendor claims. Seeking Alpha called it plainly: the AI trade broke this week on Wall Street.
The Fizzle world isn't saying AI doesn't work. It's saying the math doesn't add up yet – and that the gap between vendor story and lived experience is wide enough to move markets. South Korean stocks swinging 5% in a single session on AI sentiment is not a minor footnote. It is evidence that the financial system is starting to price in doubt.
At 44%, Spread remains the dominant narrative by source count. Deloitte, PwC, Forbes TechCouncil, Orange Business, the International AI Safety Report: the governance-and-governance-and-productivity chorus is loud and well-resourced. Worker AI access up 50%. Median SME ROI of 159%. Agentic AI crossing from pilot to production.
But here is what is structurally absent from the Spread cluster: any serious reckoning with the labour displacement evidence. The enterprise productivity reports do not mention Oracle. The governance frameworks do not engage with the HBR finding that workers are being cut based on hype rather than results. The safety report covers cybersecurity and democratic risk – it does not model wage displacement. Spread is winning the narrative by ignoring the most documented story in this week's scan.
At 12%, Autonomy ticked up from last week's 8% – driven by an arXiv paper arguing some frontier models may already constitute a form of AGI, and by TechTarget's 'so close yet so far' framing. Autonomy's series arc is the most volatile of the five worlds: it hit 29% in Scan 9, dropped to 8% in Scan 10, and is now recovering. It has never been absent. The AGI conversation is a background hum that amplifies every other signal – when Autonomy rises, Domination tends to follow within a scan or two.
One source – Inside Global Tech's coverage of criminal and state actors using AI to exploit software vulnerabilities – kept Chaos off zero. It has been near-zero since Scan 7, when it collapsed from its early series highs of 18–20%. The existential-risk narrative has not disappeared; it has been buried under more immediate concerns about jobs and returns.
Thirteen weeks ago, Domination dominated this series. Fear of AI concentration, uncontrolled power, and democratic erosion drove the conversation. Then the governance machine kicked in – safety reports, regulatory frameworks, enterprise productivity data – and Spread took the lead and held it. That is a real shift. But the jobs story was not part of that Spread recovery. It has been building separately. And now it has receipts. The question this series is quietly posing is whether Spread can hold its lead when the productivity gains its advocates cite are flowing to shareholders and executives while the displacement costs land on workers who were not in the productivity reports to begin with.
If 44% of the AI narrative is about productivity gains and only 24% is about who bears the cost of those gains – and those two groups are almost entirely different people – what exactly is the governance infrastructure that 1,000+ policy initiatives across 69 countries is designed to protect?
The articles most responsible for bending the five worlds this week.
Oracle alone cut 21,000 people this year and pointed at AI as the reason. That is not a forecast – it already happened. It turned the abstract worry about AI taking jobs into a number big enough to be impossible to dismiss.
More than half of C-suite leaders told this survey that AI adoption is actively tearing their company apart, and only 29% say they have seen serious returns on their spending. That is a striking gap between the money going in and the results coming out.
For three months in a row, more companies cited AI than any other reason when they announced job cuts. That makes this a pattern, not a one-off – and patterns are what move the overall story.
This piece makes a point most headlines miss: AI is not yet good enough to justify the job cuts being made in its name. Companies are acting on the dream of what AI might do, not what it actually does today – meaning workers are losing out to hype, not results.
AI-related stocks took a hard hit this week as investors started asking whether the enormous sums being spent will ever pay off. When Wall Street starts pricing in doubt, it changes the whole conversation about where the AI boom is actually heading.
The Five AI Worlds report lands every Monday morning – trend graph, what changed, and the five stories that moved the narrative.
Subscribe free