--- title: How the AI Story Was Told source: https://steadman.ai/newsletters/david/how-the-ai-story-was-told.html published: 2026-06-13 updated: 8th August 2026 summary: Three and a half years of AI coverage across a ten-title media spectrum (MIT Technology Review; the Financial Times and The Economist; The Guardian and the New York Times; Harvard Business Review, the two MIT Sloan titles, McKinsey and BCG), read in full text where available and charted against the AI-newsletter wave and this email's own arguments, with a switch to flip every chart between four media tiers and the individual publications. --- # How the AI story was told A news-analysis retrospective created for David Boyle's Saturday AI Thoughts. It reads the whole record of AI coverage since ChatGPT at once: 20,637 articles across a media spectrum, arranged in four tiers — technical (MIT Technology Review); business news (the Financial Times and The Economist); mainstream news (The Guardian and the New York Times); and management and consulting (Harvard Business Review, MIT Sloan Management Review, MIT Sloan's Ideas Made to Matter, McKinsey and BCG) — plus the AI-newsletter wave and the recurring arguments of this email itself. A switch flips every chart between the four tiers and the individual publications. Most titles are matched on full article text; the New York Times (headline and abstract) and MIT Sloan Management Review (headline and teaser) are metadata-only and drawn dashed, and BCG is a partial capture. **Status: Raw AI output. Not yet CEO'd (Checked, Edited and Owned).** Page updated 8th August 2026; the article counts on the page sum to 20,637 across ten publications, with coverage and matching methods differing by title (some sources carry metadata rather than full text). The data cut-off is [unverified] beyond the page's own month axis. ## I. The story that hasn't peaked Monthly AI article counts per title. A news story usually spikes and decays; this one is still climbing three and a half years on, with no peak in sight. An indexed view sets each title to 100 at ChatGPT's launch (30th November 2022) and compares growth on a log scale — a title that started near zero, like the New York Times, can grow many times over. The monthly journals run at lower volume than the dailies by design, so indexing is the fairer comparison of who accelerated. ## II. First contact The date each name (ChatGPT, Anthropic, DeepSeek, agents, and the rest) first appeared in each title, with hollow rings marking the metadata-only titles (the New York Times and MIT Sloan Management Review). Where more than one title carried a name, a line joins them; the order of arrival is the early history of the field, told by its nouns. One striking find: the FT was discussing DeepSeek in article bodies from June 2024, seven months before the January 2025 headline shock. ## III. The ideas Not products but arguments. Each idea is tracked as its share of monthly AI coverage, smoothed over three months, with the four media tiers (or individual publications) drawn against total attention, and a trajectory badge (rising, peaking, fading, stable, reversed). The ideas were discovered by a language model reading stratified samples of each corpus, then validated by counting pattern matches across every article. Concepts that matched implausibly broadly, or too narrowly to chart, were set aside (see the cutting-room floor). ## IV. The advice graveyard Ideas and advice that inflated, peaked and faded, or reversed outright, each plotted with its peak month marked. The answer to "were there bubbles of advice?" is yes, and they are dated. ## V. The AI publishing wave For each of twenty titles, the share of issue headlines mentioning AI, month by month on one shared timeline. The vertical line is ChatGPT's launch. The pioneers (Exponential View, Last Week in AI, Latent Space) span the full width; the 2023 rush titles start late; the generalist pivots turn visibly — flat, flat, flat, then vertical. Each title carries a short account of its AI journey and links to the publication itself. Two titles (Import AI, TLDR AI) were excluded for unrecovered or gap-ridden archives; titles without a meaningful pre-ChatGPT sample show their launch date instead of a baseline. ## VI. This email's territory, mapped against the press The recurring arguments of this email, each plotted against attention across the whole spectrum and classified as: the press is writing about it now, the press covered it and moved on, or white space the press has barely touched. The management and consulting tier (HBR, the MIT Sloan titles, McKinsey and BCG) is the one most likely to share this territory, so it is weighted accordingly. Each concept carries a paragraph on what the press has and has not said, drawn from a deep read of the full-text corpora. The headline finding: of twenty-five recurring arguments, only two remain genuine white space against the full ten-title spectrum — the extra hour problem, and buying intelligence by the penny. The management and consulting tier has engaged much of the territory the news titles leave untouched, though often with a different framing. The Director, Builder, Auditor team shape, for instance, is closest to HBR's "agent-manager" writing, but HBR collapses the director and auditor into one managerial role. The CEO principle (Check, Edit, Own) is nearest to HBR's accountability and governance frameworks, but those are control frameworks, not an editorial duty owned by a named human. A note added 5th September 2026: the argument this email makes has developed since these charts were first drawn. Adoption creates capacity, but capturing business value requires deliberate workflow, role and team changes. See [The vibe shift](https://steadman.ai/newsletters/david/archive.html#email-2026-09-05) and [where the argument about value now stands](https://steadman.ai/newsletters/david/ai-value-map.html); the checking framework behind it is [the CEO Gap](https://steadman.ai/newsletters/david/the-ceo-gap.html). ## VII. The cutting-room floor Every idea the analysis surfaced from all sources, with its fate: charted, too broad to mean anything, too sparse to draw, or set aside. The expandable list shows only the concepts left on the table, as a reading list of where the AI conversation went that this page did not follow. ## Method and caveats Press corpus, arranged as a media spectrum across four tiers (technical, business news, mainstream news, and management and consulting): MIT Technology Review; the Financial Times (union of the AI, OpenAI and Anthropic streams; premium-tier articles and post-June-2026 rows held to metadata); The Economist (AI topic plus a search supplement); The Guardian; the New York Times (metadata only — the Article Search API returns no bodies); Harvard Business Review (subjects "AI and machine learning" and "Generative AI"); MIT Sloan Management Review (metadata and teaser only — full text is subscription-gated); MIT Sloan's Ideas Made to Matter; McKinsey; and BCG (partial capture). Collected from paid or authorised subscriptions and open APIs for private research. Newsletter corpus: issue headlines from a newsletter-archaeology archive of twenty-eight titles, each masthead stripped before matching. Keyword matching undercounts ideas expressed obliquely. Counts are shares of monthly AI coverage, smoothed over three months. Status: Raw AI output. Not yet CEO'd (Checked, Edited and Owned).