AI Is Not “Just a Tool”

Summary

The author argues that describing AI as “just a tool” wrongly reduces a complex technology and can excuse irresponsible use or development. In the author’s framing, contemporary public discussion of AI primarily concerns LLMs, agents, commercial products, expanding data-center infrastructure and AI features added to software, rather than a single neutral instrument. The essay says a coding harness connected to a chatbot API is a paid product whose model may be retired and whose costs could rise as companies change their subsidies and pursue profitability. It also questions the provenance of local models, including who trained them, where their data came from, and what computing and environmental resources were required. As a comparison, the author describes power drills, Emacs and Vim as tools designed for specific tasks, while arguing that LLM systems can produce probabilistic mistakes, operate over longer time horizons and raise broader social and safety concerns. The essay says AI-company leaders themselves discuss severe risks, making the “just a tool” formulation inadequate for assigning responsibility. It argues that users help create demand for products whose harms are still poorly understood, while claimed benefits such as faster software development and automatically generated presentations are presented as limited. The author finally acknowledges that AI is a broader field than generative AI, but says LLMs have largely taken over the term in public discourse, contributing to its ambiguity.