GenAI for Computing Careers: A Sunny Take (cacm.acm.org)

Purdue professor Saurabh Bagchi argues for a cautiously optimistic career outlook for CS graduates in the age of generative AI: while LLMs and code-assist tools can produce simple code, they routinely generate “spaghetti” — poorly structured, fragile, and hard-to-maintain software. Because production systems still require human verification for functionality, reliability, and security, and because translating ambiguous specifications into robust designs remains a human-led task, foundational software engineering skills retain high value even as AI takes over routine coding tasks. Bagchi’s practical prescription: in college double down on software fundamentals (design, modularity, error handling), use AI tools but learn to detect and fix their mistakes, and build experience in open-source contributions (fine-tuning models, data annotation) to demonstrate integration skills. Emphasize software testing, performance optimization, and systems knowledge (algorithms, compilers, runtimes, hardware), since these areas require holistic judgment that AI can’t yet replace. The market implication is stark: mediocre developers will be marginalized, while those with deep systems and verification skills will command a premium in a winner-take-most landscape.