The information you bought is now free
Any business whose core value is “I’ll tell you how to do X” is now competing with a free interface that has read everything and will answer in 15 seconds, adjusted for your specific situation. That’s not a future risk. It’s a present reality.
Tim Ferriss published something remarkable this week: his raw BookScan numbers. His five-book catalog — all former #1 NYT and WSJ bestsellers — is tracking at roughly −57% year-over-year in 2026, following a −46% drop in 2025. The floor didn’t erode. It disappeared. ChatGPT launched in late 2022. The correlation is not subtle.
Publishers Weekly confirms the trend isn’t personal: adult nonfiction was down 9% in Q1 2026. Self-help, the most instructional of all categories, fell 26.3%. Only crafts and religion grew.
AI didn’t kill your book. It replaced the reason people bought it.
The 4-Hour Body was, functionally, a lookup table. An intelligent decision tree. In 2010 that was a valuable artifact. In 2026, an LLM that has ingested that book — and ten thousand like it — will give you a personalised protocol in under a minute, calibrated to your weight, your schedule, your specific injuries. The book doesn’t disappear. It becomes raw material the model learned from. The reader never needs to touch it.
Ferriss’s conclusion is worth sitting with: the market for information is collapsing into the chatbot. The market for transformation — for sitting with one obsessive mind, at length, on a subject it has bled for — might survive. But it will be smaller, weirder, and less commercial. Every advice-based format faces the same trajectory: how-to YouTube, prescriptive podcasts, online courses, newsletters. If your core product is transferring instructions from your head to someone else’s, the interface shift is already coming for you.
FERRISS CATALOG: PRINT SALES COLLAPSE 2022–2026
Indexed to 2022 baseline (2022 = 100) · BookScan domestic print
Source: Tim Ferriss / BookScan (tim.blog, June 2026). Values indexed from reported % changes.
US NONFICTION PRINT: Q1 2026 vs Q1 2025 BY SUBCATEGORY
Publishers Weekly · % change year-over-year
Source: Publishers Weekly Q1 2026 (cited in tim.blog, June 2026). Q1 2026 vs Q1 2025.
BITCOIN: JUNE 2026 PRICE PATH
Selected daily snapshots · USD · CoinDesk / Yahoo Finance
Source: CoinDesk, Yahoo Finance · June 2026 daily snapshots
Structured as a limited partnership with a 5-year lock-up and zero investor voting rights (except China’s national AI fund) — the deal is engineered to prevent outside interference with the founding team’s direction.
The “AI dependency paradox”: after four weeks of reliance, participants’ unassisted performance dropped below their starting baseline, while their confidence in their own ability rose.
Chair Warsh’s first meeting ended with rates unchanged; 9 of 18 members project a hike by year-end, 9 do not — Warsh himself abstained from the dot plot entirely. Nasdaq dropped 1.34% on the day.
A CFP® who retired early pushes back on the whole “you need a purpose” industry — rest, exercise, and letting your identity decompress might just be enough.
Oil is down roughly 40% from its conflict-peak as a 60-day ceasefire memo targets formal signing this week — though the IEA is already warning of a looming 2027 supply glut regardless.
Closed around $64,500 on June 17 — up ~6.4% week-over-week but down on Fed day. The Iran ceasefire optimism drove the weekly recovery; Warsh’s hawkish signals capped it.
The state added a 0.2% tax on any business activity involving digital assets — inserted last-minute into the budget. Industry groups say it’s unlikely to be reversed and sets a worrying precedent for state-level crypto taxation.
Strategy’s STRC preferred shares hit a record low, pausing the above-par sales the company uses to fund Bitcoin purchases — and forcing its first BTC sale this month to cover dividends.
Local AI models are getting genuinely good, and fast. I’ve been running models locally for a while now — it started as curiosity, mostly tinkering — but something has shifted in the last few weeks. The gap between what you can run on your own hardware and what you need a cloud subscription for is narrowing faster than I expected. That’s interesting for privacy reasons, but the more I think about it, the more it matters for anyone building on AI: the dependency on a single provider is not inevitable.
The MIT study reported this week adds a layer to this. Researchers found that participants who relied on AI to spot fake news became 21% more accurate — but after four weeks, their unassisted performance fell 15 percentage points below where they started. Confidence rose even as ability fell. There’s a name for this now: the “AI dependency paradox.” The pattern shows up in doctors, writers, data analysts. Every tool that does part of your cognitive work also atrophies the muscle it’s replacing. That’s not an argument against using AI — it’s an argument for using it more deliberately, the way a good trainer uses weights: progressively, with intention, not just handing the bar to a machine and watching.
List every product or service you pay for whose core value is “access to instructions or information.” Then ask: could a well-prompted AI replace it for free? If the answer is yes to more than half, you’ve found where your discretionary spending is quietly becoming redundant — and where someone else’s revenue is about to collapse.
Enjoyed this article?
Get the GSI Weekly Newsletter — markets, AI, and investing insights every Thursday.
Subscribe for free →No spam. Unsubscribe anytime.