Author / outlet: Nick Lichtenberg, Fortune — May 25, 2026 Type: Feature article (foundational by the borderline rule — a magazine-style synthesis citable by name for the "two phrases" framing, the BofA 0.1% anchor in context, and Mollick's KPI/consulting-arms tells).
A May 25, 2026 Fortune feature by Nick Lichtenberg frames the May 2026 AI-economy debate around two phrases that, by his account, recur at financial bubbles: "this time is different" and "nobody knows anything." The piece centers on Ethan Mollick, who has begun voicing the second publicly, and pairs his organizational critique of enterprise AI adoption with Bank of America's 0.1% estimate for AI's current lift to economy-wide productivity.
Summary of argument
Lichtenberg argues that two diagnostic phrases, both reliably present at past financial bubbles, are now appearing simultaneously and from the same speakers in the AI-economy debate:
- "This time is different" — which Templeton called "the four most costly words in the annals of investing," and which Carmen Reinhart and Kenneth Rogoff codified academically in This Time Is Different: Eight Centuries of Financial Folly (2009).
- "Nobody knows anything" — William Goldman's aphorism from Adventures in the Screen Trade (1983): "Not one person in the entire motion picture field knows for a certainty what's going to work."
By Lichtenberg's reading, the second phrase is the more telling because Ethan Mollick voiced it publicly at a New York Public Library talk to several hundred corporate leaders on a Thursday morning (May 21 or 22, 2026):
"I spend my time talking to AI labs, famous people, I talk to CEOs all the time, and nobody knows anything. We're all making this up as we go along. So anyone who's like, 'We have the playbook' — they're lying to you."
The 0.1% productivity anchor
Bank of America's own estimate for AI's current annual lift to economy-wide productivity is 0.1%, published in the same report that called AI bigger than electricity and the internet combined. Goldman Sachs (March 2026) found "no meaningful relationship between AI and productivity at the economy-wide level," while simultaneously reporting median 30% productivity boosts in the two sectors — customer support and software — where AI use has concentrated.
Lichtenberg reproduces the arithmetic behind the BofA 0.1% figure, described as the bank's own math for its own bull case:
| Term | Value |
|---|---|
| Share of workplace tasks AI can currently transform | ~20% |
| Of those, share cost-effective to automate at today's prices | ~23% |
| Labor-cost savings per automated task | ~27% |
| Labor share of total cost | ~50% |
| Theoretical productivity ceiling today | 0.66% |
| Realized economy-wide gain after friction / institutional inertia | ~0.1% |
The 0.1% figure sits below the ≤0.66%-over-10-years anchor from Acemoglu's simple macroeconomics of AI, but reaches a similar magnitude using the bulls' own arithmetic. Lichtenberg frames every serious AI-economy argument, bull or bear, as turning on whether, how fast, and at whose expense the gap from 0.1% to 0.66% to a speculated higher ceiling closes.
Mollick's "no playbook" thesis
Lichtenberg presents Mollick as supplying an organizational, rather than technological, explanation for why the productivity gap stays open. Mollick's stated framing of the central uncertainty:
"The biggest picture, there's only two questions that actually matter a lot: how good and how fast? How long does this exponential curve continue and at what point does it ease off and how sharp will it be? That determines everything else."
"There's no playbook. We're figuring it out. On one hand, that's terrifying. On the other, it's great — because that means if you create your own playbook, there's actually a source of advantage for you in that."
Mollick's jagged frontier framing, per the piece, makes naive extrapolation from "where things are today" structurally unstable.
"IT department is where AI goes to die"
Lichtenberg foregrounds Mollick's April 1, 2026 Economist "By Invitation" piece, "The IT department is where AI goes to die." Mollick's argument there is that the IT department is structurally hostile to experimentation — not from malice, but because risk-reduction mandates force experimentation onto KPI-driven paths:
"KPIs are the biggest enemy at this point. They force you into very bad paths in the experimentation phase. The very nature of saying we need a 10% improvement constrains the kind of use cases that you see."
Mollick's organizational claim is that breakthrough AI applications — the ones that replace processes rather than improve them — cannot be produced under KPI constraints. Lichtenberg presents this as the organizational manifestation of the 0.1% problem: not irrational exuberance but rational conservatism, embedded in quarterly earnings calls and performance-review cycles.
The "AI consulting arms" tell
Lichtenberg highlights Mollick's argument that the strongest evidence that "nobody knows anything" is structural rather than statistical:
"It's weird that the AI companies are all now building their own consulting arms to do AI deployment. If the models are so good that you think they're going to destroy all white-collar jobs, shouldn't they also be able to help you deploy systems?"
Per Mollick, the companies that built the technology and are most bullish on its capabilities cannot use that technology to answer the practical question of how to deploy it, and instead have to send humans. Mollick offers two readings: that the playbook genuinely does not exist yet ("nobody knows anything"), or that the same move is "the oldest playbook in the world" — a new player in town with something everyone else needs.
Reception and relation to other sources
Lichtenberg positions the "two phrases" framing alongside the "burden of proof has shifted" framing in Karma's AI-bubble piece as paired May 2026 syntheses of the bull/bear moment. The KPI and consulting-arms critiques bear on AI labor disruption, where the piece reframes "rational conservatism" as a structural brake on the displacement timelines invoked by figures such as Amodei, Doerr, and Huang. The 23%-of-20%-of-workplace-tasks math relates to AI coding agents and the enterprise AI deployment gap as a structural ceiling that coding-agent revenue would have to outrun for the bull case to convert. The BofA 0.66% today-ceiling and Acemoglu's ≤0.66%-over-10-years ceiling reach similar magnitudes over different time horizons.
Key quotes
"Nobody knows anything. We're all making this up as we go along. So anyone who's like, 'We have the playbook' — they're lying to you." — Ethan Mollick, NYPL, May 2026
"The very nature of saying we need a 10% improvement constrains the kind of use cases that you see." — Mollick on KPIs as organizational brake
"If the models are so good that you think they're going to destroy all white-collar jobs, shouldn't they also be able to help you deploy systems?" — Mollick on the consulting-arms tell
Relationships
- supports: AI Bubble Debate (Mollick KPI/consulting-arms framings are anchored bear-side structural evidence; BofA 0.1% is quantitative bear-side anchor)
- supports: Jagged Frontier (the "no playbook" frame is the organizational counterpart to the jagged-frontier capability-distribution observation)
- related: So, About That AI Bubble — Rogé Karma (The Atlantic, May 2026) (Karma's "burden of proof has shifted" sits beside Lichtenberg's "two phrases everywhere" as paired May 2026 syntheses)
- related: The Simple Macroeconomics of AI (Acemoglu's ≤0.66% / 10-year ceiling sits beside BofA's 0.66% / today ceiling — different time horizons, similar magnitudes)
- related: Ethan Mollick (primary subject of the feature)
- related: AI Labor Disruption (KPI brake reframes labor-displacement timing)
Provenance
- URL: fortune.com
- Embedded Mollick essay: Economist "By Invitation," April 1 2026, "The IT department is where AI goes to die" (economist.com)
- Embedded reference: Reinhart & Rogoff, This Time Is Different, Princeton, 2009
- Embedded reference: William Goldman, Adventures in the Screen Trade, 1983