Is AI prompting an evolution or an extinction for Bangladeshi coders?

During the AtCoder World Tour Finals held in Tokyo in July 2025, it unfolded like a high-stakes thriller from a sci-fi novel. Sixteen elite programmers, including a custom AI model built by OpenAI, hunched over identical hardware setups for a grueling 10-hour marathon. 

Their challenge was a single NP-hard optimisation problem—an evolved version of the notorious "traveling salesman" puzzle, where finding the most efficient path through a labyrinth of constraints demanded not just brute-force power but clever heuristics.  

For hours, the AI dominated, generating and testing millions of variations in parallel. Its speed was superhuman, its confidence almost unnerving. But in a late surge, Polish coder Przemysław Dębiak—better known by his handle "Psyho" and once himself an OpenAI engineer—snatched victory with 1.81 trillion points against the AI's 1.65 trillion, edging out the machine by just 9.5%.  

Barely able to speak, he managed to post on X: "Humanity has prevailed (for now!)." Later, reflecting on the contest, he admitted: "At the current state, humans—top humans, to be clear—are still much better at reasoning and solving complex problems." 

Yet he was quick to caution that AI's advantage in speed and parallelism was like "cloning a single human multiple times and working in parallel." 

That moment was electric—not just as a competition but as a metaphor. It symbolised the uneasy truth: AI is no longer just a tool in programming. It's a formidable rival.  

And nowhere does this truth resonate more poignantly than in Bangladesh, a country staking much of its economic future on its IT sector.

Bangladesh's IT sector, now a cornerstone of its $450 billion economy, employs more than a million people and exports software worth billions annually. Government officials expect more advancements in the coming years, with technology powering everything from garments to governance.

AI could supercharge this vision. Projections suggest it might boost GDP by 45% through productivity gains. The draft National AI Policy 2024 lays out plans for a National Centre of Excellence, an advisory council, and AI integration into schools and industries. 

Already, Dhaka-based firms like Intelligent Machines have rolled out AI-powered tools with remarkable results: Fordo, a precision marketing platform, delivered a 260% target stretch for Unilever, while Shobdo, a speech-recognition tool, boosted British American Tobacco's campaign accuracy by 253%. 

Yet beneath the optimism lies unease. 

For Ayesha Rahman, a junior developer fresh out of BUET, the AI shift has made her entry into the industry tougher than expected.  

"When I graduated, I thought the hardest part would be the interviews. But honestly, the hardest part now is to find companies that are hiring juniors. Many prefer to cut costs by relying on AI for repetitive coding instead of bringing in fresh talent," she said.

Still, Ayesha is trying to find solace in AI tools. She uses them daily to debug, brainstorm, and even generate sample code. "It's like a calculator for programmers," she explained. "But you still have to understand the formulas. If you don't get the logic, AI won't save you."

Her words capture the paradox: AI opens doors but also shuts them, depending on where one stands. 

Beyond the hype: Real benefits and hard trade-offs

Globally, AI has become a fixture of programming. GitHub Copilot, ChatGPT, and similar tools are now standard companions, with surveys showing that over 90% of developers used some form of AI in 2024. Thomas Dohmke, GitHub's CEO, insists AI will "reinvent" rather than replace developers. A Harvard Business Review analysis echoed this: "AI won't replace humans, but humans with AI will replace humans without AI."

In Bangladesh, AI is already at work in the ready-made garments (RMG) industry, optimising supply chains and reducing waste. Telecom firms employ AI for predictive maintenance and fraud detection, while mobile platforms like bKash use it to detect suspicious transactions.

But experts warn of pitfalls. Businesses rushing to deploy AI-generated code sometimes face crashes, vulnerabilities, or outright hallucinations—outputs that look convincing but are disastrously wrong. Prof. Feng Li notes, "Human oversight is essential. We've seen companies generate faulty code that breaks critical systems."

"AI is fantastic for boilerplate code. It can clean up a week's worth of tedious work in an afternoon. But I've also had to fix projects where AI-generated code caused security holes. The danger lies in relying too much and losing the deeper problem-solving muscles," said Tanvir Hossain, a software engineer working at a leading MFS company. 

For Tanvir, the lesson is clear: AI can accelerate, but it cannot substitute for human judgment. 

In this connection, Farzana Kabir, a senior software architect leading a team in Dhaka, remembers when outsourcing transformed Bangladesh's IT industry in the early 2000s. 

"Back then, people also feared jobs would vanish. In reality, we created new ones," she said. "But AI is different. It will take some jobs, let's not pretend otherwise. The junior roles, the routine coding—that will shrink. But at the same time, AI is also a tool. Those who learn to use it well will rise faster than before." 

She sees the future as less about replacement than redistribution. "The real competition is not between AI and humans," she explained. "It's between humans who adapt to AI and those who don't."

The skill gap and looming risks

Bangladesh's biggest challenge may be its uneven readiness. The Bangladesh Institute of Peace and Security Studies warns that "traditional skill sets are becoming obsolete," calling for urgent training in AI ethics, machine learning, and human-AI collaboration.

But reskilling takes time. Meanwhile, algorithmic bias, limited datasets, and regulatory gaps create risks. Without careful frameworks, AI could entrench injustice—spreading deepfakes, misinformation, or surveillance abuses. 

As Northeastern University Professor Upol Ehsan, originally from Bangladesh, put it, "Countries only get one true blank slate in their AI journey… Bangladesh is at that inflection point."

Ayesha worries too. "I see friends leaving for Canada or Singapore because they feel the market here isn't ready. I want to stay. But sometimes I wonder if AI is moving faster than our policies or our companies."

Between hope and hesitation

Despite the hurdles, there is momentum. Since 2010, the Bangladesh Computer Council has trained over 236,000 people—including nearly 47,000 women—in ICT and emerging fields. Universities are experimenting with AI in digital health, agriculture, and education. Telecoms and banks are embedding AI into their systems.

And for many individual coders, adaptation is already underway. Tanvir describes how he uses AI to sketch system architecture before refining it himself. "It's like having an intern who works at lightning speed," he said. "But you still need to check everything twice."

Farzana also sees opportunity in this shift. "AI is not the enemy. It's like electricity. Some people will use it to waste time, others will use it to power entire industries. Which side Bangladesh falls on is up to us."