Two trends are colliding this year.
Readers are turning away from writing that sounds like AI. And employers, especially in India, are asking whether you are fluent with AI.
It is easy to conclude that you should let AI write more. I think the opposite is true. The people who will stand out are the ones who use AI to think better and still write in their own voice.
Here is a practical guide to where AI belongs in your writing, and where it does not.
Why AI voice is a problem
On 29 October 2026, Martin Fowler published a short note on his site called LLM Voice. He describes reading a post and getting the growing feeling it came from an AI, and how his "desire to read on atrophies, sometimes to the point of revulsion."
He makes an important point. You do not need a detection tool. The reaction is human. By the time you would reach for a tool, the reader has already switched off.
The Hacker News discussion of his post drew more than 500 points and over 270 comments within a day. Readers described AI writing as sounding identical, rambling, padded and low on actual information. One commenter pointed out a familiar tell: pointless "it's not X, it's Y" constructions used to punch up something that does not need punching up.
Fowler's strongest advice is this: do not get an AI to write your first draft. Even after human editing, he says, it is very hard to remove the AI voice.
Why AI fluency still matters
At the same time, being good with AI is becoming part of your job.
Korn Ferry's Workforce 2026 Insights Report India findings found that 70% of Indian employees say AI fluency is being incorporated into employee reviews, compared with 50% globally. The survey covered 2,000 respondents in India.
But what does "fluency" mean? The Economic Times asked exactly this in a piece on how companies should measure performance when AI does part of the work. It argues that frequent AI use does not necessarily mean better performance, and that managers should look at whether employees can verify AI output, challenge questionable results and own final decisions.
The same piece cites Microsoft's 2026 Work Trend Index for India: 63% of Indian AI users named quality control of AI output as an important skill, and 59% named critical thinking.
So fluency is about judgment. Not about how many words a tool produced for you.
Where AI should touch your writing
1. Thinking before you write
Ask AI to argue against your idea. Ask what a sceptical manager would say. Ask what you have missed. This sharpens your thinking without putting words in your mouth.
2. Rough outlines
A rough structure can help when you are stuck. Fowler says outlines "may be ok, but only as far as an outline." Treat the outline as scaffolding. Change it freely, and fill it with your own sentences.
3. Research, with verification
AI can point you toward sources quickly. Always open the source and check. Fowler uses AI for writing only as a search tool, and verifies everything at the source.
4. Critique, not rewrites
Ask AI to flag unclear sentences, grammar mistakes or gaps in logic. Do not ask it to rewrite. When it suggests new wording, the AI voice slips back in. Fix the problems in your own words.
Where AI should not touch your writing
- Your opinions and conclusions. These are why anyone reads you.
- Your stories and examples. Specific details are what AI cannot invent honestly.
- Anything a person will judge you on. Your self-appraisal, your LinkedIn posts, a proposal to your manager.
- Final prose. Write the actual sentences yourself.
The second-language advantage
Many of us in India write in English as a second or third language. It is tempting to let AI smooth everything out.
Fowler addresses this directly. Many of his colleagues are not comfortable in English, and some find AI translation important. But he adds that he "would prefer rather a genuine human's clunky prose to the smooth but soulless LLM."
Your phrasing reflects how you actually think. Fix errors, yes. Do not erase yourself.
A simple routine to try
- Think out loud. Explain your point to yourself as if telling a colleague.
- Write it down as you said it. Do not polish yet.
- Ask AI for critique. "Which sentences are unclear? What is missing? Any grammar errors?"
- Fix it yourself. Use your own words.
- Check every fact at the source.
- Read it aloud once more. If it sounds like you, you are done.
Fowler calls this "Say Your Writing": speak the words in your head, write them down, then edit.
When it is fine to let AI write
There are fair exceptions. Fowler mentions a friend with Parkinson's disease for whom AI is essential to writing. He also mentions form letters and reports that mainly feed a process. For routine text no one reads closely, AI drafting is reasonable.
For anything where a human is judging your thinking, write it yourself.
The bottom line
Being good with AI does not mean letting it write in your voice. Use it to think, outline, research and check. Then write in your own words.
In appraisal season and on LinkedIn, sounding like a real person is becoming rare. That makes it valuable.
If you want to build practical AI skills that make your work sharper without losing your voice, explore our programs at The Arthi AI Collective.