"Will AI Take My Job?" — An Honest Answer for India

Part 7 of the series "AI, Without the Hype". Start at Part 1.

In most cases, no — AI will not take your job whole. It will take tasks. Almost every job is a bundle of many tasks, and AI is good at some of them and useless at others. What actually happens is quieter and more important than "robots replace humans": the person who learns to use AI well tends to replace the person who refuses to, rather than AI replacing both of them. The threat is not the machine. It is a colleague with the machine.

That is the honest answer. Now let me back it up, because reassurance without reasoning is just a pat on the head.

I build AI systems, and I have watched closely what they can and cannot do in real work. I am not going to tell you it changes nothing. I am going to tell you what actually shifts, and what you can do about it.

What past technology shifts actually did to jobs

This is not the first time a powerful new tool arrived and people feared mass unemployment. The pattern, every time, is more interesting than the fear.

When ATMs spread, people predicted the end of bank tellers. Instead their work changed — routine cash-handling shrank, and the human role moved towards advice, sales, and the cases a machine could not handle. Spreadsheets did not end accounting either; they ended hours of manual arithmetic and freed accountants to do real analysis.

The honest summary: technology tends to destroy tasks and shift roles far more than it destroys whole professions. Some jobs disappear; new ones appear that nobody could have named beforehand; and the work that remains usually moves up the ladder, towards judgement and the human parts a tool cannot reach.

AI is bigger and faster than past shifts, but the shape of the change rhymes with history. It is closer to electricity than to an asteroid.

Which tasks are most exposed

AI is strongest at tasks that are routine, language-heavy, and pattern-based. The more a task looks like that, the more exposed it is.

Most exposed:

• Routine writing — first drafts of emails, summaries, standard reports, boilerplate documents.
• Routine information work — basic data entry, sorting, simple lookups, first-line query handling.
• Predictable analysis — pulling together numbers in a familiar format, simple categorisation.
• Repetitive creative production — generic graphics, template-based design, routine copy.

Notice the recurring word: routine. If a task is the same every time, follows clear rules, and mostly moves words or numbers around, assume AI can now do a large part of it. That does not mean your job vanishes — it means that slice of your day is about to get much faster, and you want to be the one steering it.

What stays human

Now the other side, which gets far less airtime and matters far more for your career. AI is weak — genuinely, structurally weak — at:

• Judgement under uncertainty. Deciding what to do when the situation is messy, the data is incomplete, and someone has to take responsibility. AI predicts; it does not own a decision.
• Real relationships and trust. Sitting across from an anxious client, reading a room, earning someone's confidence over years. A model has no skin in the game.
• Genuine accountability. When something goes wrong, a person has to answer for it, fix it, and be trusted again. You cannot delegate that to autocomplete.
• Original strategy and taste. Knowing which problem is worth solving, and what "good" looks like for this customer, in this market.
• Physical, on-the-ground work. A vast amount of real work happens with hands, in places, with people — far from anything a language model touches.

I think about it like this. AI is a brilliant, fast, slightly overconfident intern — it produces a lot, quickly, and occasionally states a wrong fact with total confidence. You would never let an intern make the final call unchecked. The human who reviews, decides, and takes responsibility is not being replaced; that human is who the intern works for.

A practical plan to stay ahead

You do not need to become an engineer. You need to become the kind of professional who is hard to replace because you use the tool instead of competing with it. A plan that works for almost anyone:

• Learn to use AI in your own field, now. Spend a few hours a week using it for your actual work — drafting, researching, summarising. Fluency is a habit, not a course. Here is the AI skills starting point I wish I had earlier.
• Move up the value ladder. Let AI handle the routine slice, and spend the time you save on the human parts — judgement, client relationships, quality, strategy.
• Become the person who checks the work. As AI produces more first drafts, the ability to spot what is wrong, refine it, and stand behind the result becomes more valuable, not less.
• Stay curious, not anxious. The tools change every few months. The skill that lasts is the willingness to keep learning, not mastery of any one of them.

If layoffs are part of your fear, there is a real opportunity hiding inside the disruption — I have laid it out in the hidden opportunity in AI layoffs.

An India-specific note

For India, there is real reason for optimism. The economy leans heavily on services and on people, and a large young workforce that learns to use AI well does not just defend its jobs — it becomes far more productive and more competitive globally. The country is not on the sidelines either: Indian AI startups raised around $1.5 billion in the first quarter of 2026, roughly 38% of all startup funding, and home-grown players like Sarvam AI — tapped to build India's sovereign foundation model — are building seriously. That activity creates work; it does not only threaten it.

There is also a uniquely Indian edge. The hardest problems here involve many languages, code-switching between Gujarati, Hindi, English and more, and deep local context — exactly the messy, human texture that generic AI handles poorly. Building AI for that reality needs people who understand both the technology and the place. That is opportunity, not threat.

So the honest bottom line: some tasks are going to a machine, but your job is not the same thing as your tasks — and the person who pairs human judgement with a capable tool comes out ahead.

Don't fear the intern. Become the one it works for.

Next up — Part 8: AI in Indian Languages. Read the whole series at deepap.dev.