Vibe Coding: The Promise and Perils of AI-Generated Code in Software Development

Vibe coding lets anyone write software -- but comes with risks

Whether you're streaming a show, paying bills online or sending an email, each of these actions relies on computer programs that run behind the scenes. The process of writing computer programs is known as coding. Until recently, most computer code was written, at least originally, by human beings. But with the advent of generative artificial intelligence, that has begun to change. Just as you can ask ChatGPT to spin up a recipe for a favorite dish or write a sonnet in the style of Lord Byron, now you can ask generative AI tools to write computer code for you. Andrej Karpathy, an OpenAI co-founder who previously led AI efforts at Tesla, recently termed this "vibe coding." For complete beginners or nontechnical dreamers, writing code based on vibes -- feelings rather than explicitly defined information -- could feel like a superpower. You don't need to master programming languages or complex data structures. A simple natural language prompt will do the trick. How it works Vibe coding leans on standard patterns of technical language, which AI systems use to piece together original code from their training data. Any beginner can use an AI assistant such as GitHub Copilot or Cursor Chat, put in a few prompts, and let the system get to work. Here's an example:

Why vibe coding needs more than vibes for enterprise-scale solutions

Vibe coding -- which refers to using artificial intelligence (AI) and prompts to build an app or website - is all the rage these days, a trend companies and tech celebrities are getting into thanks to platforms like Cursor. But even as it makes coding as easy as sending a prompt, it presents challenges of increased scrutiny, more reviews as the codes are not production-ready and requirement of handholding for young engineers, according to industry executives. Companies are raising the bar for hiring, assessing vibe coding and introducing guardrails such as automated code review processes to address the concerns, they said. A 2025 report from GitClear that analysed 211 million lines of code written by AI copilots found that the number of duplicate codes increased to 6.6% in 2024 from 1.8% in 2023. The report highlighted that in addition to being prone to bugs (errors in coding parlance), it increases the burden on developers as they need to understand the context. Vibe coding assessments Neeti Sharma, CEO, Teamlease Digital, said while not all companies are explicitly implementing different processes, some are looking at candidates' prompt library instead of coding skills. Companies are also putting in place internal assessments to evaluate how employees are using AI. Nikunj Verma, co-founder, Cutshort, a hiring platform, said that depending on how much the company allows the employees to use AI, what it is assessing is also changing. Unlike earlier, when hiring was also about aptitude, candidates are now given problems where they can work with AI. Challenges with vibe coding Coding is among the most prominent use cases of AI. About 30% of the new codes in Google and Microsoft are written by AI, with executives saying that the numbers are likely to go up in the coming years. Despite the enthusiasm, there are multiple challenges when it comes to adopting in organisations. Nida Sahar, founder of bootstrapped cloud infrastructure platform Nife.io, said one of the biggest challenges she has been facing in the past few months has been around fresh recruits. "There is an overdependence on LLMs (large language models) and this is leading to critical thinking issues," she said. Abhishek Kaushik, cofounder, WeCP, a recruiting platform, said with AI people can build an app with just a few sentences but one needs more assessment to check if the person can build things. The company launched vibe coding assessment tools and is seeing increasing interest from companies for it. He said if the company is interviewing a senior software engineer, for instance, the assessment tool would ask the candidate to build a clone of, say X, but with five unique features, in under four hours. "If a person is able to build the system using vibe coding, it means that they are able to use AI to implement the constraints that the business needs," he said. A Bengaluru-based AI startup founder told ET on condition of anonymity that this is a concern as there is a major gap between codes written and understanding. "The number of codes written has dramatically increased but so has the burden of reviews," he said, adding that the company has increased security checks for codes and is currently putting in guardrails such as automating the code review process and practices for codes written by AI.