What 2 Years in SP Jain's MAIB Program Taught This Student

Written by Devanshi Rhea Aucharaz | Jul 14, 2026 6:45:00 PM

When people talk about AI, the conversation often revolves around the latest models, benchmarks, or breakthrough capabilities. But after two years in the Master of AI in Business (MAIB) program at SP Jain Global, Vidit K Bhatnagar realised that the biggest differentiator isn't the model you choose, it's the business question you start with.

AI is an incredibly powerful tool, but its value is determined long before the first prompt is written, or the first model is deployed. This AI Appreciation Day, Vidit reflected on how his journey shifted his mindset from building technically impressive AI systems to building solutions that solve meaningful business problems.

1. Before joining MAIB, what was your perception of AI? How has that perspective evolved during your journey in the program?

I came in as a builder, so my perception of AI was fundamentally an engineering one: better model, better outcome. I liked shipping the machinery — RAG systems, recommendation engines, multi-agent frameworks — and I measured a solution by how technically clever it was. MAIB reframed that almost completely. I stopped starting with the model. My first instinct on any problem now is to ask what decision this is meant to improve and for whom; the architecture is a downstream choice, not the starting point.

2. Can you share a project or industry experience from MAIB where identifying the right business problem was more important than choosing the AI technology?

The clearest one is JPT — the interview platform I built for SP Jain Global during the program. It's easy to treat something like this as an AI showcase: smart question generation, automated evaluation, and an impressively looking layer on top. But the business problem that actually mattered was far less glamorous — the platform had to hold up when 500+ students hit it simultaneously during a live interview window, with zero tolerance for a mid-interview crash. A failure there isn't a bug; it's the institution's credibility on the line. So the real work, and the real value, went into reliability and scale — running it on AWS EKS and hardening it across multiple workstreams — not into making the AI cleverer. That shift, from “how smart can the model be” to “what does the institution actually need this to do flawlessly,” is what made the project land.

3. In your opinion, why is asking the right business question the first step toward building an effective AI solution? Can you share an example?

Because AI amplifies whatever you point it at — including the wrong thing. A perfectly engineered solution to a poorly framed problem is still a failure, just a more expensive one. When I built a market-sizing (TAM) workbook for an AI-triaged telemedicine concept for India, the temptation was to jump straight to the triage model. The question that actually de-risked the whole idea was simpler: how large and reachable is this market, and what does a patient genuinely need at the moment of triage? Once you answer that, the technical choices largely make themselves.

4. How has MAIB helped you connect AI with business strategy, customer needs, or organisational decision-making instead of viewing AI purely as a technical tool?

MAIB deliberately places the technical and the strategic side by side. In the same stretch, I did a deep dive into a frontier model's architecture (DeepSeek's V3) and strategic cases on Netflix's India play and the Lovable platform. That combination trains a specific muscle — you learn to move from “how does this work” to “why would a business or customer pay for it” in the same breath. My work outside class reinforces it daily: I build an AI product for an edtech platform serving working professionals, so every feature is judged by whether it helps a learner or an institution, not by how elegant the pipeline is.

5. AI is often seen as a replacement for human work. Where do you believe human judgment, creativity, or domain expertise remain indispensable?

In framing, judgment calls, and knowing when the output is simply wrong. AI is excellent at generating options; it is poor at deciding which option fits a messy, real-world context full of constraints it cannot see. When I build course content or full curricula with AI, the model drafts fast — but a human still has to decide what a UAE working professional actually needs, catch factual and tone errors, and make the taste-level calls. The same holds in strategy, negotiation, and product. AI widens the search space; someone with domain context still has to choose from it. AI raises the floor — expertise still sets the ceiling.

6. As AI becomes more integrated into businesses, what responsibilities do future AI and business leaders have to ensure its ethical, responsible, and impactful use?

The first responsibility is honesty about limits — not overselling what a system can do, especially in high-stakes areas like health, finance, and education. Beyond that: keep a human accountable for consequential decisions, be transparent about where AI is used, protect people's data, and design so the technology augments people rather than quietly displacing their judgment. As builders, we set defaults that scale to thousands of users at once, so getting those defaults right — on privacy, fairness, and safety — isn't optional. It's the job.

7. If you could share one key takeaway with aspiring students or young professionals about learning AI today, what would it be?

Learn to build the model and interrogate the problem — but if you only have time to master one, master the questions. Tools change every few months; the ability to figure out what is genuinely worth solving doesn't go out of date. Build real things, ship them, and stay ruthless about “why does this actually matter to someone?” Behind every smart AI solution is a better business question — start there.

About the author

Vidit K Bhatnagar is a Master of AI in Business (MAIB) student at SP Jain Global’s Dubai campus.