Why LLMs Make Learning to Code More Important, Not Less (senthil.learntosolveit.com)

Senthil Kumaran, a seasoned software developer at Uber, delivered a compelling talk at the OMSCS Conference 2026, asserting that the rise of large language models (LLMs) does not signal the decline of programming skills but rather emphasizes their importance. Kumaran contends that while LLMs enable more accessible coding solutions, true programming—rooted in understanding and intuition—remains critical. He argues that the notion of "coding is dead" is misleading as programming continues to evolve, supported by significant investment in LLM-based tools, which enhance rather than replace the coding skill set. Kumaran highlights that tools like LLMs enable learners to tackle previously daunting programming resources more efficiently, facilitating deeper understanding of complex concepts through creation rather than rote learning. This shift allows developers to focus on the essence of problem-solving and foundational knowledge, crucial for developing robust software. He advocates for a learning approach that leverages LLMs as teaching aides, guiding learners to discover solutions step-by-step. This strategy not only democratizes access to programming knowledge but also reinforces the necessity of a solid grounding in programming fundamentals as the landscape of technology continues to evolve.