Why You Should Almost Never Use AI for Writing

When you hand the pen to a machine, you trade the messy, intentional labor of thought for a polished but hollow output.

What the Argument Is Saying

Eric Grunewald’s recent essay, “Why you should almost never use AI,” makes a bold claim: for most writing tasks, you should avoid AI altogether. The piece has sparked a lively discussion on Hacker News, gathering over two hundred points and a hundred‑plus comments. While the article itself is not reproduced here, its central thesis is clear—relying on AI to generate text is, in most cases, a poor choice.

Why the Cost Often Outweighs the Convenience

Even without a detailed list of reasons from the source, several practical concerns emerge when you consider the trade‑offs of AI‑generated prose.

• Skill erosion. Writing is a muscle; the more you let a tool do the work, the weaker your own ability becomes.
• Surface‑level thinking. AI models excel at pattern matching, not at deep, original synthesis. The result can feel familiar but lack true insight.
• Ownership ambiguity. When the text is produced by an algorithm, the line between your ideas and the model’s training data blurs, raising questions about authorship and responsibility.
• Contextual blind spots. AI lacks lived experience. It may miss cultural nuances, ethical implications, or the subtle tone you intend.
• Long‑term knowledge decay. Relying on AI to fill gaps can prevent you from internalizing information, leading to a fragile grasp of the subject matter.

Preserving Creative Integrity

If you value work that reflects your unique voice, you need safeguards that keep the human element front and center. Below are concrete habits to cultivate.

1. Draft first, edit later. Write a raw version yourself before consulting any tool. This ensures the core idea originates from you.
2. Use AI as a mirror, not a source. Feed your own draft into an AI to see how it would rephrase the same thoughts; compare and keep what feels authentic.
3. Set strict boundaries. Define which stages—if any—AI may touch. For example, allow it to suggest headings but not to generate paragraphs.
4. Document the process. Keep a log of where AI was used. This transparency helps you track dependence and maintain accountability.
5. Practice regular reflection. After each writing project, ask yourself what you learned, what you struggled with, and whether AI helped or hindered your growth.

When AI Might Still Have a Role

Grunewald’s stance does not imply that AI is universally useless. There are narrow scenarios where the tool can be a pragmatic ally, provided you remain vigilant.

• Technical scaffolding. Generating boilerplate code comments or standard legal clauses can free mental bandwidth for higher‑order tasks.
• Language translation checks. Using AI to spot glaring errors in a translation you’ve already produced can be a safety net, not a primary author.
• Accessibility support. Converting text to simpler language for readers with disabilities can be a compassionate use, as long as you verify accuracy.

In each case, the human author stays in control, using AI as a low‑risk helper rather than a creative engine.

Building a Sustainable Human‑AI Collaboration

The conversation sparked by Grunewald’s article points to a broader cultural question: how do we integrate powerful tools without surrendering the very qualities that make our writing valuable? The answer lies in intentional design of the workflow.

Start by defining the purpose of each piece of writing. If the goal is to explore a novel argument, the draft must be yours. If the goal is to polish an already solid argument, AI can serve as a second pair of eyes. Treat the model as a collaborator with a narrow, well‑scoped role, and always perform a manual, critical review before publishing.

Finally, remember that the durability of knowledge depends on the depth of your engagement. When you wrestle with ideas, you build mental models that persist long after the document is archived. Letting AI shortcut that wrestle may produce a fleetingly attractive paragraph, but it will not nurture lasting understanding.

Choosing to write without AI most of the time is not a rejection of technology; it is a commitment to preserving the craftsmanship, accountability, and intellectual resilience that define meaningful communication.