Hype Is a Business Tool

Scott Jenson argues that the current frenzy around large language models (LLMs) is primarily a business tool driven by hype, not an accurate reflection of capability. He insists on calling the technology LLMs rather than the catch-all "AI" and traces the pattern to the Gartner Hype Cycle - innovation trigger, peak of inflated expectations, trough of disillusionment, slope of enlightenment, plateau of productivity. Historical examples include MOOCs (Udacity, Coursera) and the early mobile/web era (WAP phones, Siemens devices, Nokia, the slow maturation to the iPhone 3GS), showing how high expectations collapse before real, incremental progress emerges.

The core claim is that venture funding and PR inflate unrealistic promises to extend runway, citing firms from Theranos to OpenAI as examples of hype tactics. Operational costs for LLMs currently far exceed revenues, so many projects fail to deliver; a trough of disillusionment is likely before sensible applications surface. The practical prescription is a UX-grounded, bottom-up approach: focus on small, solvable problems, understand LLMs’ strengths and limits, and build iteratively rather than relying on grandiose claims. A follow-up will examine how anthropomorphizing LLMs distorts how people use them.