Hype Is a Business Tool (jenson.org)

This piece argues that the current frenzy around “AI” is better framed as a hype-driven business cycle around LLMs (Large Language Models) rather than a settled technological revolution. The author warns that boosters, vendors and investors are inflating expectations—often blaming user prompts when systems fail—while the underlying tech remains young and error-prone. Using the Gartner Hype Cycle as the lens, they trace familiar rises-and-falls (MOOCs, early mobile) to show how new tech typically hits a Peak of Inflated Expectations, falls into a Trough of Disillusionment, and only later finds practical, productive uses. For AI/ML practitioners the takeaway is practical: LLMs can do impressive things, but their deployment is costly (model compute and serving often outstrip revenue), many enterprise pilots underperform, and naive product assumptions persist. The author urges a UX-driven, bottom-up approach—solve small, boring problems, measure real user outcomes, and avoid anthropomorphizing models—so teams can climb the Slope of Enlightenment. Expect a likely correction in the short term, not necessarily doom, and prioritize engineering and UX rigor over hype if you want sustainable, useful LLM applications.