If your company just made AI use expected, you are not alone. The question is not whether to open the tool. The question is how to learn it fast enough that it helps your work, and how to show real outcomes when someone starts counting usage.
Why this is in the news
In early November 2026, Indian outlets including Times of India, Economic Times, CNBC TV18 and Mint covered an open letter circulating among current and former Microsoft employees. The group signs as "Concerned Microsoft Workers." According to reporting first published by The New York Times, they say that for many staff, "working at Microsoft has become joyless and demoralizing."
They describe being pushed to use AI tools across their jobs with little or no training, then spending time correcting weak output. The letter asks Microsoft to involve employees in how AI is built and used, and to stop linking AI use to rewards or reprimands. The same reporting describes a "speed-at-all-costs" approach.
That letter began circulating before Microsoft's October 26 Voluntary Exit Offer to about 60,000 eligible employees in India, the US, the UK and Canada. Microsoft's chief people officer has said the buyouts are not related to how AI is changing work. Treat the letter and the exit offer as two stories in the same news cycle, not as proof that one caused the other. If you are weighing an exit offer yourself, start with our voluntary exit guide.
The Microsoft story is the hook. The rest of this guide is for you, if your Indian workplace just made AI mandatory and skipped the teaching.
My take
The problem is not mandatory AI. The problem is mandatory AI without teaching, then scoring people on how often they use a tool.
Counting prompts measures compliance. It does not measure better decisions, safer delivery or clearer writing. When people fear looking behind, they stop raising risks. That is AI theatre.
I teach AI to working professionals in India, one to one. Earlier roles at Zoho, AstraZeneca and Pfizer taught me the same lesson from inside large organisations: tools without practice become policy theatre. Expect fluency. Teach it. Judge the work.
If your next review will ask about AI, also read how to prepare for an AI fluency review.
What to ask your manager this week
Send a short note. Keep it practical.
- Which AI tools are approved for my role, and which are not?
- What data must never go into an AI tool, including personal consumer accounts?
- How will AI fluency be assessed in my appraisal: usage, outcomes, or both?
- Is there protected time for training, or am I expected to learn after hours?
- Who reviews AI-assisted work before it reaches a client or production?
If the answer to question 3 is only "use it more," offer outcomes for your role. Examples:
- Program manager: time to produce a weekly status pack, with fewer follow-up questions
- Analyst: hours saved on a repeatable report, with the same or lower error rate
- Marketer: draft-to-publish time, with a human edit checklist
- Support lead: first-draft reply time, with escalation accuracy unchanged
Show outcomes, not usage counts
Keep a one-page log. Update it every Friday.
| Date | Task | AI help used | What changed | What I caught |
|---|---|---|---|---|
| 7 Nov | Weekly status email | First draft only | 40 minutes to 15 | Invented a date I fixed |
| 8 Nov | Meeting notes | Summary then rewrite | Same day share | Missed an action owner |
Bring three completed rows to your review. Add one example of work you refused to put into AI because of confidentiality. That last point often matters more than people expect.
Also watch how your writing sounds. If everything suddenly reads generic, fix it before your manager notices. Use AI without sounding like AI and use ChatGPT without switching off your brain are useful companions here.
A two-week self-training plan
You do not need a new degree. You need deliberate practice on work you already own.
Days 1 to 3: one task, one tool. Pick a recurring task. Write your current steps. Ask AI for a first draft only. You edit everything. Save the before and after.
Days 4 to 7: add a second task and a red line. Make a "do not paste" list: passwords, OTPs, client contracts, Aadhaar or PAN details, medical or payroll data, private messages. Use sample data when you practise.
Days 8 to 10: make the AI critique you. Before you send work to a human, ask the tool what looks weak, invented or incomplete. Note every hallucination. That habit is the real skill.
Days 11 to 14: package one proof point. Write a half-page note for your manager: task, baseline, new process, result, risk you avoided. Ask for feedback on the outcome measure, not on how many chats you opened.
If you want a longer runway, use the 90-day AI roadmap for non-IT professionals or how to learn AI as a non-IT professional in India. For tool choice once the skill is clear, see which AI tool you should pay for.
When to push back
Learning is not optional in most knowledge jobs now. Blind compliance still is optional.
Push back when:
- Your score depends on usage volume with no quality bar
- You are told to put confidential work into an unapproved personal tool
- There is no time to verify output before it ships
- Raising a risk gets labelled as "not AI-ready"
- The only training is a login link and a dashboard
How to push back without sounding difficult:
"I am ready to use the approved tools. I need clarity on confidential data rules, time to verify output, and an outcomes measure for my role instead of a usage count."
That is a professional request. It protects the company as much as it protects you.
For the wider jobs picture in India, read AI and jobs in India in 2026.
Frequently asked questions
My company made AI mandatory. Where do I start?
Start with one recurring task, not every tool. Ask how AI fluency will be judged, keep a weekly outcomes log, and practise for two weeks on work you already do.
Is tracking how often I use AI a fair appraisal metric?
Usage alone is weak. It rewards clicks. Ask to be judged on outcomes, verification, error rates and the judgment to leave sensitive work out of AI tools.
What should I ask my manager after an AI mandate?
Ask which tools are approved, what must stay out, how fluency will be scored, and whether training time is protected. Offer role-specific outcomes instead of usage counts.
When should I push back on AI at work?
Push back when usage is scored with no quality bar, confidential work is pushed into personal tools, or there is no time to verify before shipping.
Does the Microsoft workers letter mean AI mandates are wrong?
No. The reported Concerned Microsoft Workers letter describes pressure to use AI with little training and asks the company to stop linking AI use to rewards or reprimands. For Indian workplaces, the useful lesson is teach first and measure work, not clicks.
Make AI help the work, not the dashboard
An AI mandate can either become a usage contest or a skills upgrade. You get more of the second when you ask better questions, keep proof of outcomes, train on real tasks and refuse fake metrics.
At The Arthi AI Collective, I help professionals turn AI into results they can show at work. Explore the programs. Teams that want structured rollout support can also look at corporate AI training.
Arthi Rajendran, The Arthi AI Collective