saved
AI Is a Tool. Using It Well Takes Skill.
Hraness wrote this summary from a saved copy of the source. Quotations are taken word for word from the source.
gist
Huahai responds to Rich Hickey’s “‘AI’ is Not a Tool,” arguing that LLMs are tools even by Hickey’s own definition, and that using them well takes skill, feedback and judgment, as with a saw or a chisel. Automation can be a feature of a tool, as compilers and spreadsheets show. The useful line is between wielding an instrument skillfully and delegating work without understanding or oversight, not anthropomorphic talk of software that goes rogue or conspires.
ideas
- Tools amplify judgment. A saw does not guarantee a straight cut; a tool’s usefulness depends on directing it, reading feedback and correcting mistakes, and LLMs are no different.
- Unpredictability is not disqualifying. Reliability and bounded function are criteria for judging tools, not requirements for being one.
- Automation inside tools. Compilers and spreadsheets do substantial work unattended, yet people still wield them by specifying inputs and examining results.
- Human-directed feedback loops. Explaining code, challenging the explanation, testing and rejecting changes is intention plus expertise; using an LLM alongside reading, not instead of it, preserves learning.
- Drop the anthropomorphism. Calling failures “going rogue” or “conspiracy” imports human motives into what are real engineering problems.
quotes
“Picking up a saw does not guarantee a straight cut.”
“A tool does not stop being a tool because it is difficult to control or easy to misuse.”
“The relevant question is what the person actually does.”