When I studied Electrical Engineering, one of the subjects that stayed with me was the 8085 microprocessor. With its limited set of registers, even simple programs—addition, subtraction, finding the smallest or largest number—could feel like daunting exercises. But those exercises taught more than assembly language and microprocessor fundamentals. They taught efficiency: how to use limited resources carefully and deliberately.
Years later, while building mobile applications, the devices were far more powerful. Yet performance still mattered. To avoid noisy-neighbour-like situations and deliver a smooth, consistent, buttery user experience, platforms created programming models and marketplace rules. If an app failed to comply, it might not be published—or worse, the operating system could terminate it.
Then came the cloud era. In many ways, efficiency became increasingly influenced by the scale, automation, and pricing models offered by cloud providers. As organizations expanded their cloud usage, optimizing resource consumption and controlling costs emerged as a specialized field, leading to the growth of the cloud optimization industry.
And now we are in the age of AI. Many consider this a pivotal moment in the evolution of technology, marked by rapid advances in artificial intelligence and its growing integration into everyday tools and workflows. Whatever perspective one takes, AI is clearly changing how people engage with technology and expanding what digital systems can do.
A story I came across online—so take it with a pinch of salt—captures the absurdity well. Someone gifted a mobile phone to their parents. The phone had replaced the traditional power button with an AI agent. When the parent wanted to switch off the phone, pressing the button launched the AI agent instead. So the parent simply told the agent to switch off the phone. The agent then opened the power-off screen, making it easier for the parent to complete the action.
Convenient? Maybe. Efficient? Not really. That interaction likely consumed (if not processed locally) tokens, API calls, compute, energy, and possibly even water—just to do something that a simple physical button once handled instantly.
That, to me, is the real question. Are we making technology smarter, or are we designing systems that make people depend on unnecessary layers of intelligence for simple tasks?
My interpretation is simple: AI may be getting more powerful, but if we are not careful, people may become less thoughtful about the cost of convenience.
tendulkar-uvāca 