# Judge a Technology by Shipped Value Not the Funding Narrative ## Claim A technology's real worth is revealed by the durable problems it provably solves, not by the funding narrative or branding built around it, so the right evaluation question is "show me the working product," not "how big is the promise." <!-- atomicity: 1 idea ✓ --> ## Scope & conditions - Holds during investor-driven hype cycles where incentives reward inflated promises and where the audience evaluating the claim lacks the expertise to check it directly (the same asymmetry that makes hype possible in the first place). - Requires distinguishing "no product yet" (genuine early-stage research, real future value) from "no value possible" (pure narrative) — the test targets narrative-substituting-for-proof, not all unproven work. - Breaks where a technology is explicitly pre-product and honestly framed as research, not sold as a shipped capability. ## Evidence A skeptical-pragmatist keynote on the AI hype cycle contrasts two poles: chatbot products reporting user-engagement metrics because they can't show profit, versus DeepMind's AlphaFold, which computed and published roughly 200 million protein fold predictions for free and led directly to new medicines. The talk's throughline is that "artificial intelligence" as a label does rhetorical work — leaning on science fiction and human pattern-perception — that the technology's actual track record doesn't independently support; the corrective is to look past the name to what has actually shipped and been verified. ## Links - [[3. Resources/Permanent Notes/A Cynical AI Bet Only Needs Bosses to Believe]] — extends: both describe how a narrative can substitute for evidence when the audience evaluating an AI claim can't verify it directly. - [[3. Resources/Permanent Notes/What a Recommendation Omits Identifies Its Real Selection Criterion]] — applies-in: both are cheap, pre-analysis filters — check the omission, or check for a working product — that can end a bad evaluation before the deep dive starts. - [[3. Resources/MOCs/AI Cost & Tokenomics MOC]] — applies-in: separating funded narrative from shipped value is a load-bearing filter for this MOC's economics tracking. - [[3. Resources/MOCs/Evidence & Claims Literacy MOC]] — applies-in: Joins the cluster on testing hype against checkable evidence. ## Tests / falsifiers - For any AI product claim, check whether a specific, independently reproducible output exists (a published dataset, a benchmark, a shipped feature) versus only a roadmap or investor deck. - Track whether firms citing user-engagement as their headline metric eventually convert that engagement into a specific, verifiable capability, or whether the metric substitutes for one indefinitely. ## Maintenance - Last tended: 2026-08-16 - Next review: 2026-11-16 ## Source - [[3. Resources/Readwise/Articles/Keynote After the AI Hype – What's Real, and What's Next]] — the AlphaFold-vs-engagement-metric contrast and the "AI is a fundraising name" thesis.