Let's start with the truth: we understand why you feel this way.
You've watched ChatGPT write in ten seconds the query your teacher spent an hour explaining. You've seen reels telling you coding is dead. Your seniors say placements are luck anyway. Your phone buzzes every few minutes with something more fun than a pivot table. And somewhere in your head, a quiet voice keeps saying:
- “Fees toh bhar di hai, certificate mil hi jayega.”
- “Class miss ki toh kya, recording hai na.”
- “Assignment? ChatGPT se 2 minute ka kaam hai.”
- “Job ke time dekh lenge. AI toh hai na.”
If you've thought any of these, you're not lazy and you're not stupid. You're a normal student in 2026, living inside the most distracting device ever invented, being told by the internet that effort is optional.
But here's what nobody on Instagram tells you. The shortcut works right up until the day it doesn't. And that day usually has an interviewer sitting across from you.
What actually happens in a data analyst interview now
A composite story, based on the kind of interview our students and faculty describe regularly.
Rohan's resume looked great. Data Analytics course, completed. SQL, Excel, Python, Power BI, all listed. Four projects on GitHub. He'd built every one of them with ChatGPT: copy, paste, run, screenshot, done.
The interviewer smiled, shared her screen and pasted a query.
Interviewer: “One of our freshers got this from an AI tool. It runs fine. Our finance team says the revenue number is too high. Can you tell me why?”
The query on her screen: “Total 2025 revenue by city”
SELECT c.city,
SUM(o.order_total) AS revenue
FROM orders o
JOIN customers c ON c.id = o.customer_id
JOIN order_items i ON i.order_id = o.id
WHERE o.order_date >= DATE '2025-01-01'
GROUP BY c.city;Two mistakes, and neither one shows an error.
- Joining
order_itemsrepeats each order once for every item in it. An order with 3 items gets counted 3 times, so revenue is inflated. - There's no end date, so 2026 orders are sneaking into “2025” revenue.
The habit that catches it: count the rows before and after every join. If 1,000 orders turn into 3,200 rows, something is being counted more than once.
A student who has done the joins assignment spots this in under a minute. Rohan couldn't, because he had never actually written a join himself.
The fix: drop the join you don't need, and close the date range
SELECT c.city,
SUM(o.order_total) AS revenue
FROM orders o
JOIN customers c ON c.id = o.customer_id
WHERE o.order_date >= DATE '2025-01-01'
AND o.order_date < DATE '2026-01-01'
GROUP BY c.city;Rohan stared at it. He knew what JOIN meant, sort of. He'd never had to think about one. The room went quiet in that very specific way you never forget.
That's the part the “AI toh hai na” crowd misses. Companies know you have ChatGPT. Everyone has ChatGPT. That's exactly why they don't test whether you can use it. They test whether you can think without it, and whether you can catch it when it's wrong.
The data backs this up. PwC analysed over a billion job ads for its 2026 AI Jobs Barometer. In US job ads, the entry-level roles most exposed to AI were seven times more likely than the least exposed ones to ask for traditionally senior skills, like judgement, leadership and strategic decision-making. (PwC, 2026) Freshers are still being hired. But the fresher who can only press buttons is not.
Is AI killing data analyst jobs? What the numbers say
Some jobs, yes. Just not the ones you think.
Which jobs grow, and which shrink (US, 2025–2035)
Read it like this. The jobs AI is eating are the copy-paste jobs: data entry, routine reports, work where you follow the same steps every day. US projections show office and admin work losing about 752,000 jobs by 2035. The jobs growing fastest are the ones where people think with data. Data scientists alone are expected to grow nearly ten times faster than the average job.
Here's the uncomfortable twist. If you “learn” data analytics by copying AI answers and never practising, you're training yourself for the copy-paste job. The exact one that's disappearing.
The same pattern shows up worldwide, and it's even stronger in India. The World Economic Forum surveyed over 1,000 employers. They expect big data specialists to be the fastest-growing job to 2030, and they rank analytical thinking as the number one skill, with seven out of ten companies calling it essential. Employers in India expect data analyst and data scientist roles to grow 54% by 2030, against 41% globally. (WEF, 2025)
And NITI Aayog says India's tech sector could lose 1.5 million jobs, or create up to 4 million new ones, by 2031. Which way it goes depends on whether people actually build skills. (NITI Aayog, 2025) Right now, India's supply of AI talent is only about half of what companies need, according to NASSCOM figures in the same report. The jobs are there. The prepared people aren't.
AI makes weak students weaker, and strong students faster
Think about a calculator. It didn't kill maths. But a student who doesn't understand maths will happily type 50 × 20, hit a wrong key, get 10,000, and write it down without blinking. A student who understands maths looks at 10,000 and thinks, “That can't be right.”
AI works the same way, just with much bigger consequences.
Researchers from Harvard Business School and Boston Consulting Group tested this with 758 BCG consultants. When the task was something AI handles well, people using AI finished 12% more tasks, 25% faster, and at higher quality. But on a tricky task that looked easy and wasn't, people using AI were 19 percentage points more likely to get it wrong than people working without it. They trusted a confident answer they couldn't check. (Dell'Acqua et al.)
And AI is wrong more often than reels suggest. In DataSpace, a 2026 benchmark of 410 realistic data-analysis tasks, even the best of six frontier AI models got about one in three wrong. (DataSpace, 2026)
AI makes a skilled person faster. It makes an unskilled person confidently wrong.
Now the good news. In the same study, on tasks AI handled well, the consultants who started below average gained the most. In the 2023 working paper, their scores rose 43%, against 17% for the top performers. So if you're not a “topper”, AI can be your biggest opportunity. But only if you know the basics well enough to tell when it's helping and when it's guessing.
What are SQL, Excel, Python and Power BI for, now that AI exists?
When the IMF studied 2024 US job ads for data analysts, the top skills were data analysis, SQL, communication, Python, Power BI, Tableau, dashboards, problem-solving and data visualisation. (IMF, 2026) Employers didn't drop them because AI arrived. Here's why, in plain words.
SQL
AI can write the query.
You have to know if it's right. AI has never seen your company's database. It guesses. You saw above what one wrong guess looks like.
Excel
AI can write the formula.
You have to notice when it quietly breaks after someone adds 200 rows, and explain to your manager why two reports show different totals. That's the part of the job that pays.
Python
AI can generate the code.
You have to read it. AI-written Python looks equally professional whether it's right or completely wrong. If you can't read it, you're just hoping.
Power BI or Tableau
AI can build a basic dashboard.
You have to decide what the dashboard should measure, set up the data model so numbers don't double-count, and explain the story to someone who doesn't read charts. Making pretty charts alone is no longer a job.
Notice the pattern? In every tool, AI does the typing. You do the thinking. And thinking is exactly the skill you only get by practising, making mistakes, and fixing them yourself.
Want to see how we teach each one? SQL · Advanced Excel · Python · Power BI · Tableau, or all of them together in our Data Analyst course in Nagpur.
How to use AI while you learn, without letting it think for you
We don't want you to stop using AI. We want you to use it the way a good analyst does. Here's the routine we recommend for every assignment:
- Try it yourself first. Write the query, formula or code on your own, even if it's wrong. Those ten minutes of struggle are where the learning happens.
- Then ask AI. Share your attempt and ask it to review it, or ask for its own version.
- Compare line by line. Where are the two different, and why? If you can't explain a line of AI's answer in your own words, you're not done.
- Check the result against something you trust. A row count, a total you already know, three rows worked out by hand. Never submit a number you haven't checked.
A prompt that works well for learners: “Don't give me the answer. Tell me which line of my query is wrong and give me a hint.”
This is how working analysts use AI too. The typing gets faster. The responsibility for the number stays with you.
Let's talk about your phone
We're not going to tell you to delete Instagram. We use it too. But do this quick calculation once, honestly.
That's more time than most students spend practising in their entire course. You don't need to give it all up. Take back even a quarter of it and you'll be ahead of most of the people you'll compete with in interviews.
A few things that genuinely work for our students:
- Phone in another room while you practise. Not face-down on the desk. Another room.
- Notifications off during class. One buzz costs you far more than the few seconds it takes to check.
- 30 to 45 minutes of practice every day beats a four-hour panic session before the exam.
A letter from your teacher
I'll be honest with you, because I think you deserve honesty more than a sales pitch.
Some days teaching is frustrating. I prepare a class, and half the batch is missing. The half that's there is scrolling under the desk. Assignments come in with the same ChatGPT answer, same variable names, same mistakes. And I think: yeh bacche apne hi future ke saath kya kar rahe hain?
But I also remember the students who struggled, stayed after class, got things wrong ten times and then finally understood. They weren't always the smartest in the room. They were the ones who kept showing up. And I've watched those students walk into interviews calm, because they'd already solved harder problems in class.
You or your parents paid the fee. Please don't let it buy you only a certificate. The certificate opens the door. What you practised decides whether you stay in the room.
Use AI. Seriously, use it. I use it every day. But use it like a calculator in the hands of someone who knows maths, not like a crutch you can't walk without.
Come to class. Do the assignment even when it's boring. Especially when it's boring. The version of you who sits in that interview six months from now will thank you.
Saurabh JoshiCTO, Unisoft Technologies, Nagpur
What a good course should give you, and what only you can give
A course is a partnership. Neither side can do the other's half.
What a good institute should give you
- Practice on real, messy data, not just clean textbook tables
- Training on using AI tools and checking what they produce
- Projects you can explain line by line in an interview
- Faculty who have fixed wrong numbers in real work
- Mock interviews with live questions
What only you can give
- Showing up to class, even on tired days
- Doing assignments yourself before asking AI
- Asking questions when you're lost, not scrolling
- Daily practice, even 30 minutes
- Treating the fee as an investment, not a ticket
Ask any institute you're considering, including us, how they cover the left column. And be honest with yourself about the right one.
Still unsure? Let's talk honestly.
Come and talk to our counsellors, or bring your doubts straight to a faculty member. Ask us the hard questions: how AI fits into the syllabus, what practice looks like, what kind of role is realistic for your background. If data analytics isn't right for you, we'll tell you.
Or see what's inside our Data Analyst course in Nagpur.
Unisoft Technologies, Dharampeth, Nagpur. Training students since 2000, with 80,000+ trained, a 4.6★ Google rating and an authorised Pearson VUE test centre. Email: mail@unisoftindia.org
Questions students ask us
Will AI replace data analysts?
No. AI is taking over routine, copy-paste data work, but demand for people who can think with data is growing. Employers in India expect data analyst and data scientist roles to grow 54% by 2030 (World Economic Forum), and the US Bureau of Labor Statistics projects data scientist jobs to grow 34.6% from 2025 to 2035, against 3.5% for all jobs. The bar for freshers is higher, though: you need to be able to check and fix what AI produces.
If AI can write SQL and Python, why should I learn them?
Because jobs and interviews test whether you can think with data, not whether you can type a prompt. AI-written queries often run without errors and still give wrong answers, and only someone who understands SQL can spot that. Learning the basics is what lets you use AI safely, and faster.
Do companies allow ChatGPT in data analyst interviews?
Don't count on it. Many first rounds are live SQL or Excel tests on the interviewer's screen, and some companies explicitly don't allow AI assistants. Even where AI is allowed, you'll be asked to explain the output and fix it when it's wrong, which you can only do if you've practised.
How should I use ChatGPT while learning SQL?
Write your own query first, then ask ChatGPT to review it or show its version. Compare the two line by line, and check the result against something you trust, like a row count or a total you already know. A useful prompt is: “Don't give me the answer. Tell me which line of my query is wrong and give me a hint.”
Can I get a data analyst job with just a certificate?
A certificate helps your resume get noticed, but most interviews include a practical round. PwC's 2026 research found that in US job ads, the entry-level jobs most exposed to AI are seven times more likely than the least exposed ones to ask for traditionally senior skills, such as judgement and decision-making. What gets you hired is solving problems in front of the interviewer.
I'm from a commerce or arts background. Can I become a data analyst?
Yes. Business sense matters as much as coding in most analyst roles, so start with Excel and SQL. One of our students, Tejaswini Ambilkar, came from a commerce background, mastered SQL in our Data Analytics track and was placed as an SQL developer at Lighthouse Info Systems in Nagpur.
I get distracted easily. Can I still learn data analytics?
Yes. Almost every student struggles with focus now. Keep your phone in another room while you practise, switch notifications off in class, and practise for 30 to 45 minutes every day. Small daily effort beats big last-minute effort.
About the author
Saurabh Joshi
CTO, Unisoft Technologies, Nagpur · 16+ years in databases, analytics and web technologies
Saurabh leads technology at Unisoft Technologies and teaches SQL, PL/SQL, Oracle DBA, Advanced Excel and Tableau in its Data Analyst program. Students he has trained now work as SQL developers and database administrators, including at Lighthouse Info Systems in Nagpur. He also runs the institute's own data and web operations, so he uses AI tools every day, and checks their output the way this article asks you to.
- OCP DBA
- OCA DBA
- Oracle SQL Associate
- Excel Expert (MO-211)
- Excel Associate (MO-210)
Keep reading
All articles on the Unisoft blog →
Sources
- U.S. Bureau of Labor Statistics, Employment Projections 2025–2035, news release USDL-26-1422 (27 August 2026). https://www.bls.gov/news.release/pdf/ecopro.pdf
- World Economic Forum, Future of Jobs Report 2025 (January 2025), including the India economy profile. https://reports.weforum.org/docs/WEF_Future_of_Jobs_Report_2025.pdf
- PwC, Two futures for jobs in an AI era: 2026 Global AI Jobs Barometer (June 2026). https://www.pwc.com/gx/en/issues/artificial-intelligence/job-barometer/2026/2026-global-ai-jobs-barometer-global-findings.pdf
- Dell'Acqua et al., Navigating the Jagged Technological Frontier, Organization Science (March 2026), https://doi.org/10.1287/orsc.2025.21838. First released as Harvard Business School Working Paper 24-013 (September 2023), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4573321
- Li et al., DataSpace: Benchmarking Data Agents for Verifiable Analytics over Heterogeneous Workspaces, arXiv preprint (August 2026). https://arxiv.org/abs/2608.03451
- IMF, Bridging Skill Gaps for the Future: New Jobs Creation in the AI Age, Staff Discussion Note SDN/2026/001 (January 2026). https://www.imf.org/-/media/files/publications/sdn/2026/english/sdnea2026001.pdf
- NITI Aayog, Roadmap for Job Creation in the AI Economy (October 2025). https://www.niti.gov.in/sites/default/files/2025-10/Roadmap_for_Job_Creation_in_the_AI_Economy.pdf