AI is cutting jobs at record pace – but who's actually getting richer?

The gap that defines everything

The Domination–Spread gap is the single number that tells you most about where the AI narrative stands. At its worst, back in scan 2 (14 April), Domination led Spread by 24 percentage points – the discourse was dominated by fear, concentration of power, and harm. By scan 10 (28 June), Spread had opened a 22-point lead in the other direction. That is a 46-point swing across the series. This week, scan 12, Spread leads by 18 points (44% vs 26%). The gap has narrowed slightly for the second consecutive scan – from –22 to –20 to –18. That is not a reversal. But it is a direction of travel worth watching.

Spread remains the dominant narrative for the fifth scan in a row. That run – from scan 8 through scan 12 – is the longest sustained stretch of optimistic framing in the series. Enterprise AI reports, governance frameworks, and leadership summits all keep reinforcing the idea that AI is a powerful tool humans can steer. Deloitte, PwC, Forbes, and Orange Business all filed that line this week. The institutional discourse is settled on the Spread story.

Domination is climbing back

But this week's Domination reading – 26% – is the first uptick since scan 6. It had fallen from 38% (scan 2) all the way to 24% (scans 9, 10, and 11). Now it has ticked up two points. One week is noise. Two consecutive upticks would be a signal. The driver is unmistakable: AI-cited layoffs. TechCrunch's running tracker puts the 2026 figure at roughly 120,000 workers, with 56% of layoff events explicitly naming AI as the cause per SkillSyncer's live data. Oracle's 21,000 cuts alone generated significant narrative weight. CNBC reported that AI is now the leading reason companies give for cutting jobs. These are not speculative stories about a future threat. They are current, documented, named-company events.

What makes this week's Domination cluster structurally interesting is that the concentration-of-power story and the labour-displacement story are now fusing. AI Weekly's editor documented frontier labs funding each other in closed loops. Bending Spoons filtered 800,000 applicants down to 286 hires. The same week, three humanoid robotics companies moved toward IPO. The pattern: productivity gains are real, and they are accruing to a very small number of entities.

Fizzle (AI-Fizzle) quietly contradicts the Spread consensus

Fizzle sits at 12% this week – down from 17% last scan and its lowest reading since scan 9. But the individual sources are punchy. Terminal X cited S&P Global data showing 42% of companies abandoned most AI projects in 2025. Writer.com found only 29% of enterprises report significant ROI despite massive investment, and 54% of C-suite executives say AI adoption is tearing their organisations apart. Harvard Business Review went further, arguing that layoffs are being driven by AI hype and executive speculation rather than proven AI performance.

This creates a direct contradiction with the Spread consensus that the enterprise discourse is settling. If half of AI initiatives fail to deliver and a majority of executives report internal organisational damage, the 'powerful tool delivering measurable gains' framing is at best incomplete. The structural absence in this week's coverage is any serious attempt to reconcile those two readings. The optimistic reports and the failure data sit in parallel, rarely cited together.

Autonomy (Singularia) and Chaos (Paperclipalypse): both rising

Autonomy climbed from 12% (scan 11) to 14% this week – driven by a cluster of AGI-focused content including a Google DeepMind benchmarking paper and an arXiv paper noting that Anthropic's president has suggested Claude may have already reached 'a kind of AGI'. That claim – if it lands in mainstream coverage – would be the single biggest narrative shift of the series. It hasn't yet. But the academic and research layer is moving in that direction.

Chaos doubled, from 2% to 5% – the highest reading since scan 4. The Future of Life Institute's AI Safety Index explicitly warned of 'recursive self-improvement' and labs racing toward capabilities without credible safety plans. Anti-AI protests hit OpenAI, Anthropic, Google, and a16z on the same day. These are still minority signals. But the series low for Chaos was scan 9, at 0%. It has now risen in three of the last four scans.

The plain-language read across the full series

Thirteen weeks ago, the discourse was dominated by fear. Domination and Chaos together held more than 45% of the narrative in the earliest scans. Then the enterprise wave hit – productivity reports, governance frameworks, summits – and Spread took over, peaking at 46% in scan 10. That shift was real. The question this week is whether it was a ceiling. Layoff data is accumulating faster than governance frameworks can respond to it. The gap between 'AI as powerful tool' and 'AI as mechanism for concentrating wealth' is not a philosophical debate anymore. It is showing up in quarterly earnings reports and workforce statistics simultaneously.

What is missing

Almost no coverage this week asked who owns the productivity gains once jobs are gone. The governance discourse is sophisticated on transparency, explainability, and compliance. It is nearly silent on distribution. That is the structural gap the chart is not yet measuring – and probably should be.

If Domination closes to within 10 points of Spread in the next two scans, the optimistic consensus that has defined the series since June will need to be reassessed. Watch the layoff tracker.