Bengal said No to Computers - I won't say No to AI

August 01, 2026

I stumbled upon an uncomfortable question while working on a demo about property-based testing. For years, I have been an advocate of Jqwik, a remarkable open-sourced library that implements property-based testing in Java. Well before the AI revolution, I had been talking on and off stage about it. Recently, I discovered that Jqwik’s latest version carries an anti-AI clause in its usage policy.

Is it unethical to use AI? I don’t think the answer is simple. I battle with two sides of the argument at once: the author is right to be concerned while I can’t afford to agree with him.

My non-negotiable - my bottom line, the reason I do anything - is to put food on the table for my family. Part of that means doing what the industry expects - to stay relevant, for one thing, to keep up with the new speed of doing things, for another. If AI makes all of our jobs irrelevant tomorrow, I’ll gladly become a farmer, clean homes - even steal bread if I have to. Providing for my family is not a philosophical position, it is as real as it gets. It is survival. I do not have the luxury of opting out of what the industry demands.

That does not however mean that I would stifle the work of others - far from it. I would like to believe that we can mutually respect each other while disagreeing on specific topics. I have deep respect for the author’s work and his stance on AI. That is why, in my experiments with Jqwik, I have only used version 1.9, the last release before the anti-AI clause was added.

But the irony of it all is in why I am playing with Jqwik at all: to catch what AI misses.

Jqwik is one of the few tools we have to catch AI’s faults. It is a testing library that lets you define invariants and properties that are validated at runtime. Unlike handwritten tests, which sample test cases, Jqwik explores the problem space for inconsistencies. It was always a powerful tool but today it is invaluable. The ability to interrogate what we build solves many problems that AI brings, ranging from hallucinations to subtle logical gaps.

Doesn’t that highlight the power of human ingenuity? A library written entirely by hand that catches AI’s mistakes? Ironically, if the tools we need are the ones that discourage interfacing with AI, what are we left to correct AI with?

You have heard of Bangalore. But you probably have not heard of Kolkata, a city in Bengal that was once one of the most prominent cities in India. There was a saying I grew up with: what Bengal thinks today, India thinks tomorrow. That is no longer true. Today, no one thinks of Bengal at all.

In the 70s, West Bengal’s communist government resisted computerization of its industries, banks and schools. While the rest of India partook in the digital revolution, Bengal resisted change. The state and the city fell behind by decades, if not a century, fading into oblivion today.

The politicians of my homeland made a choice that generations of young adults are still paying for today. What could have been a prosperous IT hub of the east is now a dystopian land embroiled in corruption and brain drain. The 70s government resisted change. It was a principled stance that was wrong and devastatingly costly.

I don’t want to - nor have the luxury to - make the same mistake of resisting change. For better or for worse, I am learning how to engineer software alongside AI. Refusing the industry’s trend risks my bottomline of providing for my family. It is a risk I am not willing to gamble on, and neither should you.

See anything to improve? Edit this post!