Resistance is futile: AI is now writing code

Why this concern over AI-assisted coding?

A Harvard Business Review study found a 20% drop in software development roles over two years as AI tools like GitHub Copilot, ChatGPT, Cursor, Tabnine and Devin streamline coding, reduce costs and modernize systems. AI automates tasks, detects errors and optimizes workflows, shifting developers from manual coding to overseeing AI-driven processes. ‘Vibe coding’, which leverages large language models (LLMs) to generate, debug and refine code, further enhances efficiency. While AI-assisted coding boosts developer productivity, it is also replacing routine coding jobs and lowering real wages of coders.

Also read | Surviving GenAI: Indian IT recodes its future as a decades-old model crumbles

Does this mean AI will kill manual coding?

OpenAI CEO Sam Altman predicts AI agents will become virtual co-workers, handling complex software engineering tasks. InMobi CEO Naveen Tewari warns that 80% of the company’s coding will be automated by the year-end, leading to job losses within two years. OpenAI chief product officer Kevin Weil believes that “this is the year that AI gets better than humans at programming forever,” while Zerodha CTO Kailash Nadh expects junior developers to be most affected. Nvidia CEO Jensen Huang has also cautioned that AI threatens IT jobs, even suggesting that children no longer need to learn coding.

Does everyone agree with this prognosis?

Andrew Ng calls discouraging coding due to AI automation “the worst career advice ever,” arguing AI tools make coding more accessible. Gartner sees AI boosting senior developers’ productivity, AI agents automating tasks and AI-driven software boosting demand for AI engineers. Bill Gates insists AI lacks adaptability and precision, keeping humans essential.

Also read | Now, build a 3D game on the fly, courtesy GenAI

Do AI-assisted code have any limitations?

Simon Willison, co-creator of Django, advises never trusting a code until you’ve seen it work—or fail. Deloitte warns AI-generated lines of code pose quality, security and integration risks, requiring thorough reviews. They struggle with niche languages, outdated security checks and intellectual property concerns. Integration issues arise when AI-generated lines of code conflict with existing architectures, coding standards, and overly complex code further complicate maintenance. Deloitte recommends strict oversight.

How should we prepare ourselves?

Anthropic CEO Dario Amodei predicts AI will surpass humans in most tasks within two years. Gartner says 80% of software engineers must upskill by 2027. But AI still lacks creativity, leadership and ethics. Hence, as AI-assisted coding and automation grow, so the roles in governance, collaboration and business strategy will thrive. Former Google CEO Eric Schmidt urges entrepreneurs to use AI to quickly create prototypes of their ideas. The future belongs to those who embrace AI, not resist it.

Also read | A Google GenAI expert weighs in on why companies are clamouring for AI agents