Vibe coding is here to stay — but engineers and computer scientists say 'learn to code' is still good advice

The landscape of software engineering is undergoing a transformation with the emergence of 'vibe coding,' a term that describes the use of natural language to instruct AI tools to generate code. Kevin Wu, CEO of the AI startup Leaping AI, observes that while many are embracing this trend, not everyone is effectively utilizing it. He recounted a troubling experience with a new hire who struggled to understand the AI-generated code, highlighting a significant gap in foundational knowledge. Vibe coding has sparked debate in the tech community regarding the traditional advice to 'learn to code.' Experts from major tech companies and computer science academia emphasize that while AI can enhance coding efficiency and enjoyment, it cannot replace the expertise of trained software engineers. Nenad Medvidovic, chair of the Computer Science Department at USC, likened the situation to placing an inexperienced driver behind the wheel of a race car, warning that a lack of understanding can lead to critical errors. Research from Stanford University indicates that the rise of AI is already impacting entry-level coding positions. Their study found that among younger workers in AI-heavy fields, those aged 22 to 25 experienced a 13% decline in employment opportunities. For early-career software developers, employment fell nearly 20% from its peak in late 2022 through mid-2025. Todd Millstein, a computer science professor at UCLA, likens coding with AI to collaborating with a junior engineer. He explains that while AI can assist in generating code, the responsibility for understanding and reviewing the output still lies with the human coder. Engineers must often step in to handle complex tasks that AI struggles to grasp fully. Millstein also pointed out that initial code creation is just one part of a larger software lifecycle that involves maintenance, bug fixes, and updates. In high-stakes sectors like finance and healthcare, having a clear understanding of code is crucial for ensuring security and proper functionality. Jigar Bhati, an engineer at OpenAI, underscores that AI tools serve more as assistants rather than replacements for human coders. He notes that while vibe coding can facilitate rapid prototyping, a successful product relies on the expertise of seasoned engineers who understand the intricacies of the system. Ritvika Nagula, a software engineer at Microsoft, stresses that a solid grasp of coding principles is essential for effectively utilizing AI tools, as the quality of AI responses is heavily influenced by the prompts given. If context is lacking, the AI may produce irrelevant solutions. Despite its limitations, vibe coding is recognized for making programming faster and more enjoyable. Antara Dave, a product designer at Microsoft, points out that it allows for quick experimentation and the generation of innovative ideas. As AI technology continues to advance, Millstein and Medvidovic highlight the importance of adapting educational curricula to prepare students for the evolving landscape of software engineering. They assert that while the fundamentals of coding will remain relevant, integrating AI tools responsibly into the learning process is essential. Ultimately, the role of software engineers is likely to evolve, incorporating vibe coding into their skill set without diminishing the need for foundational knowledge.