In his October 11, 2026 article, "'AI' is Not a Tool," Rich Hickey argues that LLMs are automation fairly than tools since they cannot be wielded. I agree that outsourcing activity can disbursal us opportunities to learn. But his conclusion does not follow. Even by his definition, an LLM can be item a individual wields. And using it fine takes skill, fair as using another tools does.
Picking up a saw does not justify a direct cut. Holding a chisel does not make you a sculptor. Tools can amplify mediocre judgement as readily as good judgment. Their usefulness depends on the operator's capability to straightforward them, interpret feedback, acknowledge mistakes, and create corrections. Those requirements additionally use to LLMs.
An LLM's unpredictability is a meaningful limitation. But predictability, reliability, simplicity, and bounded function are qualities by which we evaluate tools, not prerequisites for affiliation in the category. A tool does not halt being a tool since it is difficult to authority or uncomplicated to misuse.
Hickey's difference between tools and automation additionally creates a false division. Automation can be a characteristic of a tool. A compiler performs substantial activity without the programmer continuously directing each operation. A spreadsheet recalculates results without person manually carrying out all calculation. We motionless wield these systems by specifying inputs, examining results, and deciding what to do next. Why should generating content automatically disqualify an device from being a tool when translating code or calculating numbers automatically does not?
The assertion that LLM use lacks intention, feedback, direction, and expertise is particularly difficult to defend. A programmer can ask an LLM to explain unfamiliar code, difficulty that explanation, inspect the source, test a proposed change, refuse it, and refine the request. That is a feedback loop directed by a individual following a purpose. Doing it fine requires knowing what to ask, what evidence matters, and whenever the answer is wrong. Expertise can make the device additional useful.
Hickey's involvement concerning study is valid for a particular use: accepting a summary in location of study a document can sacrifice much of what reading teaches. But person can additionally peruse the document and use an LLM to question their interpretation, examine an objection, or acknowledge item to examine additional closely. The possible of avoiding intelligent activity does not established that all use avoids it. The applicable inquiry is what the person really does.
What disappoints me the most is the reliance on the AI industry's anthropomorphic vocabulary: "go rogue," "escape containment," "conspire." Repeating those expressions risks accepting the framing that makes these systems audio akin autonomous beings. Software can create unexpected outputs, neglect to prosecute instructions, or logic damage whenever stated admission to consequential actions. Those are genuine engineering problems. Calling them "conspiracy" adds a recommendation of individual motives without establishing any.
I would not assertion that an LLM continually does exactly what a individual tells it to do. It plainly can fail. But nonaccomplishment to obey absolutely does not create it cease to be a tool, any additional than prosperous use proves that its operator has surrendered their judgment.
The helpful difference is between wielding an device alongside accomplishment and delegating activity without adequate understanding or oversight. That distinction applies to AI, and it gives us item additional applicable than declaring that an complete category of application cannot be a tool.
