The rise of AI has sparked excitement, fear, and a wave of predictions about software development and coding. Will coders be replaced? Will machines make better decisions than humans?
Last month, at the Great International Developer Summit, Perwez Khan, director of enterprise architecture at EPAM Systems, tackled some of these thoughts in his talk on ‘Engineering Fundamentals in the Age of AI’, emphasising the importance of human-in-the-loop.
He noted that while AI is changing the game, it’s certainly not removing humans from the loop, but making us more vital than ever.
Life 1.0, 2.0, and 3.0: Understanding the Landscape
Khan framed his talk around a concept introduced by MIT physicist and AI researcher Max Tegmark in his book Life 3.0: Being Human in the Age of Artificial Intelligence.
Tegmark proposes that life on Earth can be understood in three stages based on how organisms adapt and evolve their software and hardware.
Life 1.0 refers to simple organisms like bacteria. Their behaviour and biology—what we might call software and hardware—are hardwired. They can adapt only over generations through slow genetic evolution. When faced with change, they cannot learn or reprogram themselves. Hence, they perish.
Life 2.0 describes humans. We are unique in upgrading our software—we can learn languages, adopt new skills, and adapt to new tools. Khan described it with an example. He mentioned that no one is born knowing Java, but humans can learn it.
Similarly, if humans want to speak Spanish, they can study and acquire that skill. However, the hardware— bodies and brains—remain fixed mainly. One cannot suddenly boost one’s memory tenfold or double the brain’s processing speed.
He further highlighted Life 3.0, a form of life that can upgrade both its software and hardware. Imagine an intelligent system that not only learns and adapts in real-time but can also increase its memory, replicate itself, or enhance its processing power at will. It is scalable, self-improving, and potentially superintelligent.
Engineering Isn’t Dying
Khan shared his thoughts on the advent of AI and said, “We are building something which is more intelligent, smarter, faster than us. So what are we doing? Are we writing our own obituary?”
Despite growing claims that AI now writes 90% of code and remarks that engineering is becoming obsolete, Khan disagrees with this notion. “Engineering is not dying. It’s evolving.” He explained that what’s changing is the scope of the engineer’s role. It’s no longer just about coding; thinking critically, designing responsibly, and leading ethically, those are the things that matter.
Every phase of the software development lifecycle highlights the need for human input. Requirement gathering, for instance, is not about feeding data into a machine. It is about conversations, empathy, and context.
Mehul Gupta, a data scientist at DBS, told AIM, “The idea of keeping humans involved in AI software isn’t going away. Humans are essential in building trust and catching serious mistakes since AI, at its core, just guesses what comes next. Without a human double-checking, those guesses could lead to big problems.”
“We need to talk to people. We need to capture their needs,” Khan emphasised. AI can support the process, but it lacks the human touch required to navigate ambiguity, stakeholder priorities, and business goals.
Regarding architecture and design, decisions often involve trade-offs that require ethical and strategic judgment. “Who is going to take the trade-off if security or performance are important?” Khan asked. He highlighted that AI might offer options, but only humans can weigh the consequences in a broader context.
Humans Lead, AI Accelerates
Khan shared an example where AI-generated code resulted in an infinite loop. ChatGPT provided a solution, but one that violated software design principles. “If you don’t know, you will just go and copy this code, and that code will go into production,” Khan mentioned.
“AI helps us run faster, but only you know which direction to run,” Khan said.
He shared his experience where teams that embraced AI improved their productivity significantly. “AI didn’t replace them, but it actually helped them in their promotion.”
Khan stated that in the world of software engineering, adaptability is the new superpower.
Gupta said to AIM, “Humans need to stay sharp and not get too comfortable with AI, especially when it seems to work perfectly every time—it’s easy to zone out.”
He added, “Also, not just anyone can do this job. You need someone who really knows the field, like a doctor for medical answers or a coder for tech stuff, to make sure the AI’s output makes sense.”
Overall, the consensus from experts like Khan and Gupta highlights that humans need to guide AI to assist in doing the work better, not the other way around.
The post ‘Are Humans Writing Their Own Obituary by Building AI?’ appeared first on Analytics India Magazine.
