‘90% of what programmers write is boilerplate, AI is doing a great job eliminating….’: Zoho’s …

Posted on Author AIJobNexus

This post was originally published on this site.

Zoho founder Sridhar Vembu believes AI can eliminate up to 90% of programmers’ “boilerplate” code, but maintains that human insight remains crucial for tackling the “essential complexity” in software development.

“When people say “AI will write 90% of the code” I readily agree because 90% of what programmers write is “boilerplate” he wrote in a post on X (formerly Twitter).

Vembu highlighted a concept from the classic tech book The Mythical Man-Month, distinguishing between “essential complexity” —the core logic that requires human creativity—and “accidental complexity” —the repetitive tasks AI can handle.

Advertisment

“AI is doing a great job eliminating the accidental complexity,” he wrote while stressing that humans must still manage the core problem-solving aspect of programming.

Vembu further posed the question of whether AI could discover entirely new patterns, calling it a rare skill demanding “taste” or “knowing where to dig” — traits that might not be easy to replicate through brute force.

“In essence, AI can make mincemeat of patterns already discovered (by humans). Can it find totally new patterns? As with humans, that is much rarer and a quality known as “taste” or “knowing where to dig” or “follow a hunch or conviction all the way” is needed to discover new patterns. I don’t know if AI can do this. I don’t know if that can be brute forced,” Vembu added.

Earlier this year, Sridhar Vembu stepped down as CEO of Zoho to assume a new position as chief scientist, focusing on research and development (R&D) initiatives. Following this, co-founder Shailesh Kumar Davey became the company’s new CEO.

Netizens reaction

Vembu’s post, which has crossed over 79,000 views, sparked a lively debate among developers and tech enthusiasts.

“AI will magnify the gap between developers who can design systems well and those who merely translate requirements to code. The former will become more valuable and powerful, the latter will find themselves automated. This widening gulf between “idea multipliers” and “implementers” will shape the software profession,” a user wrote.

“Won’t it require a lot of experience of maybe having coded without AI to be able to handle essential complexity? I feel the importance of having a deep understanding of the fundamentals of systems will be very critical. How will those who start programming in the AI era develop this?” another wrote.

“It’s like transitioning from C to Visual Basic ..” a third wrote.Â