Introduction: AI is not magic, it is a machine learning from data

Artificial Intelligence (AI) is reshaping Software Development and research. From code generation to protein folding, the hype is everywhere. Auto-generated code, intelligent testing, accelerated drug discovery; the tools, products and start-ups are arriving fast. But one thing is often missing from the conversation - clarity.

AI is not a silver bullet. It cannot sense our intentions or solve organisational challenges without context. AI is not a substitute for framing a problem, architecture, or design. Treating AI as a fast-learning, capable colleague helps shift the focus from AI as a superpower to AI as a smart assistant.

The foundations still matter

However tempting it might be to quickly involve AI, we must ask fundamental question: What problem are we trying to solve?

AI cannot define business goals, prioritise features, or manage trade-offs. These continue to be the responsibility of analysts and architects. Together, they ground a solution and shape meaningful boundaries for innovation.

Without the resilience from architecture and relevance from analysis, AI generates noise, not solutions.

Adopting AI without structure results in expensive failures. AI must stand on a foundation of domain knowledge, human judgement, and thoughtful design.

Design Thinking: More important than ever
The belief that AI may replace Design Thinking is a fundamental error. Design Thinking is more vital than ever. Its foundation is Empathy: understanding user needs before building features.

AI can support creativity but only when the challenge is well-framed.

In the true Design Thinking pattern — Empathise, Define, Ideate, Prototype, Test — at each stage, AI plays a supporting role but not a leading one.

Teach AI — just like a human

Contrary to the thought that AI “just knows,” it “knows” because it can predict patterns very quickly based on vast training data. However, it does not “understand” anything. Knowing and understanding are two different things.

No AI is trained on any company’s vision, user pain points, or the team’s Definition of Done. Expecting AI to “just do it” is like expecting a junior Engineer to deliver a production system without any onboarding.

Retraining AI using our own context, boundaries, and relevant examples will bring AI closer to solving the problem statement.

Use AI to accelerate, not to initiate!

Once the challenge is well-defined AI can generate code, write test cases, draft documentation, and summarise research papers.

This leads to faster development cycles and less human error, although human in the loop is important.

Without clarity, speed turns into chaos.

In our journey of AI adoption, we must place our trust in Engineers and draw on their experience. Bold visions and KPIs from the boardroom are good and inspiring, yet the real progress comes from a practical and collaborative approach by redefining the achievable and measurable goals underpinning the overall vision.

Delivering small value and learning from feedback are the best ways to build trust and momentum. As we continue, we must also remember that AI itself is evolving alongside us.

AI in experimental design and scientific discovery

AI is revolutionising experimental design and scientific discovery by detecting patterns in large datasets.

Most powerful example is DeepMind’s AlphaFold. By solving the challenge of protein folding with deep learning, it is transforming biology and medicine saving years of research.

AI in collaboration and knowledge sharing

R&D also requires effective sharing of the results. AI can read thousands of research papers, patents, and documents to find connections, suggest collaborations, and highlight unseen opportunities.

This breaks down silos between disciplines and teams. For organisations, this means, greater collaboration, efficient knowledge reuse, and cross-functional innovation.

Bring people along at their own pace

AI adoption is not just a technical challenge; it is an emotional one. People are excited, anxious, and even afraid of their future.

Leaders must create psychological safety by introducing AI gradually. Reassure teams that AI is here to augment, not replace, emphasising the impending need for upskilling.

Software Engineering fundamentals still matter in the age of Artificial Intelligence

History has shown that industrial revolutions often begin with fear before becoming widespread and accepted. When electricity was first introduced, it felt magical as well as dangerous. Today, it is a necessity. Likewise, AI too could unlock a whole new set of possibilities that were once unimagined.

Not every problem needs AI

Some problems are too small, too ambiguous, or too human for AI to solve. That is not a weakness — it is a signal for wisdom.

Experienced leaders recognise when not to rely on AI. We should avoid applying AI in situations where human reasoning, ethical judgment, or contextual expertise are more appropriate.

Vibe Coding: Powerful but with the right foundations

Vibe Coding refers to a style of programming where developers describe their intentions in natural language and AI generate the code.

Vibe Coding is gaining traction. But these are not shortcuts to mastery.

Without a grounding in Software Engineering, AI will simply reproduce poor patterns — faster.

Tune out the panic. Tune into structure

I purposely crafted this section heading in the true AI style! and I mean it.
Despite the benefits and availability of choices, we do not need to learn hundreds of AI tools quickly and in panic; rather, we should start with a structure by asking ourselves:

• Where are the inefficiencies in our workflow?
• Where can AI reduce repetitive toil?
• Where is accuracy and trust most critical?

Do not copy competitors — they might be wrong

Responding to competition is natural, but following competitors without understanding can lead us in the wrong direction.

We should align AI with our core strategy, our people, and our customers’ needs. While learning from others is important, following blindly is like walking a path without knowing where it leads.

If we copy a failing competitor, we risk inheriting their failure too.

AI Digital Transformation?

AI is not just a tool. It feels like the next wave of Digital Transformation in making — like Cloud, DevOps, or Mobile in the past.

But this wave must be anchored in values. The hype will pass but our values and principles should not.

Conclusion: Software Development, reinvented but not replaced

AI will help us reinvent Software, but Engineering will always lead the way.
Instead of using AI everywhere, use it where it makes sense and produces real value. Use it to free up time, reduce routine, and accelerate learning and innovation.