Use AI to build the first prototype, but don’t stop learning the fundamentals: Intel’s Aditi Nanda

There was a time when becoming a developer meant memorizing things that now seem increasingly unnecessary. A new programming language, a framework, a library, and the cycle never really stopped. Learn one, become comfortable with it, and then start preparing for whatever comes next.

AI is beginning to disrupt that cycle. But according to Aditi Nanda, Director of Ecosystem and Industry at Intel, the answer is not for developers to stop learning. It is to become more selective about what they learn.

Speaking at DevSparks Chennai, Nanda said AI is taking over much of the rote work involved in software development, but that does not make engineering fundamentals less important. If anything, it makes them more valuable, as developers are increasingly responsible for deciding what AI should build and whether what it produces is fit for use.

Nanda compared the change to the arrival of GPS. Drivers no longer need to memorize every road, but they still need to know when to accelerate, brake or turn around when the road ahead is blocked.

“You don't have to remember the syntax. You don't have to remember the frameworks. But you have to decide what the outcome is,” she said.

Also Read

Why AI leaders are divided over slowing down artificial intelligence

Stop memorizing, not understanding

For developers, that could mean spending less time trying to master every new framework or programming language and more time understanding the systems those tools are being used to build.

Nanda pointed to a familiar problem with generative AI: it can get a team from an idea to a prototype remarkably quickly, but getting that prototype into production requires a different set of skills.

“Use AI to get to the first prototype, the demo, but don't stop learning your basics. Don't stop learning your fundamentals of engineering,” she said.

Those fundamentals extend beyond writing code. Nanda emphasized the importance of understanding the market and the user’s problem before deciding on a technical solution.

“You need to understand what the pain point of the market or the consumer or the user is,” she said. “We can create the best solutions, but if nobody is going to use it, it's a waste.”

System knowledge is another area she believes developers cannot afford to lose. As AI applications move between cloud, PCs, edge devices and data centers, developers need to understand where a workload should actually run. For an enterprise application, that decision can involve cost, security, and compliance.

“Sensitive HR or financial information, for instance, may not be appropriate to send outside an organization's environment. Architecture therefore remains a developer's responsibility even when AI is writing parts of the implementation,” she said.

Also Read

DevSparks Chennai returns: Explore AI for the physical world, manufacturing, and deeptech

The professional is more than an AI user

That distinction between a developer who understands AI and an AI-dependent developer becomes particularly important as AI makes software development accessible to people who may not have traditional skills.

Nanda gave the example of using Cursor to quickly build an app while waiting for a flight. Creating a functional-looking application is now possible in a fraction of the time it once took. But getting that application ready for millions of users is a different challenge.

“A professional will be responsible enough to understand how a user is going to look at this. What are the bugs? How can I verify it? What is the compliance? What is the cost? Where do I store the data? How is the security? They will inquire about all of these,” she said.

For Nanda, the developer's future therefore rests on three broad capabilities: understanding the ‘why’, understanding the ‘how’, and being able to take a solution into the real world.

The first is understanding the user and the problem. The second is system and engineering knowledge, including architecture and the trade-offs between cost, effectiveness and time to market. The third is execution: verification, reliability and making sure the product works when it reaches users.

She said that while AI can augment each of these capabilities, it cannot remove the responsibility from the person building the product.

“Use AI as a tool. Use AI as an augmenting engine. Don’t let it do the thinking for you,” Nanda said.

For developers navigating an industry where a framework can become obsolete before people have mastered it, that may be the more useful lesson: the skills worth holding on to are increasingly the ones that remain valuable even when the tools change.