Why a Microsoft Engineer Still Bets on Humans Over AI

Since joining Microsoft in 2018, Devikar has navigated the rapid integration of artificial intelligence into the software development lifecycle. Working on projects like the Microsoft 365 Agents SDK, he found that the most significant time savings occur not during initial code generation, but during the iterative refinement process. Tasks that once required weeks of manual adjustment now move to completion in days, allowing teams to pivot from frantic deadline-chasing to deeper product optimization.

However, the ease of automation creates a new technical hurdle: the "hallucination" problem. Devikar warns that relying on AI requires a deep understanding of the underlying logic, as developers must be capable of auditing outputs that can appear correct while containing subtle errors. He cautions against over-reliance, suggesting that engineers who stop developing their foundational knowledge risk contributing to a surge in low-quality "AI slop."

Despite the industry’s ongoing shift, Devikar remains skeptical that AI will trigger a mass displacement of software engineers. He views the current technological surge as analogous to the rise of the internet—a disruptive force destined to become mundane infrastructure. He maintains that while AI can suggest approaches or draft code, the responsibility for architectural decisions and product strategy remains firmly with human experts who prioritize continuous learning.