AI is a tool and using it well takes skill

The way you can use AI assistants to help with your work fundamentally depends on your skill in directing them.

Rich Hickey argued in an October 11, 2026 article that Large Language Models (LLMs) are automation, not tools, because they cannot be "wielded." Huahai, writing for yyhh.org on October 12, 2026, counters this, asserting that LLMs can indeed be wielded and that using them effectively requires skill, similar to other tools like saws or chisels. Huahai argues that predictability, reliability, simplicity, and bounded function are qualities by which tools are evaluated, not prerequisites for being classified as a tool. The author criticizes the AI industry's anthropomorphic language, such as "go rogue" or "conspire," stating these terms incorrectly suggest human motives for software failures. Instead, software's unexpected outputs or failures to follow instructions are engineering problems. The core argument is that the distinction lies not in whether a system is a tool or automation, but in whether a human wields it with skill or delegates work without adequate understanding.

The debate over whether AI, specifically LLMs, are tools or automation hinges on the definition of 'wielding.' While some argue LLMs cannot be controlled like traditional tools, this perspective overlooks how users interact with complex systems like compilers or spreadsheets. These existing automations are still considered tools because users direct their inputs, interpret outputs, and make decisions. Similarly, LLMs can be wielded through directed questioning, feedback loops, and critical evaluation of their responses, requiring operator skill rather than passive consumption.

This framing challenges the notion that LLMs are inherently uncontrollable or fundamentally different from other advanced technologies. It suggests that the true differentiator is the user's expertise in leveraging these systems, rather than the systems' inherent characteristics. The reliance on anthropomorphic AI terminology is criticized for obscuring practical engineering challenges and promoting a false sense of independent agency in software, which can mislead users about accountability and control.