Jim Nielsen discusses how large language models (LLMs) shift the paradigm of software development by allowing computers to infer user intent, reducing the need for highly detailed coding. He warns that as LLMs become more prevalent, distinguishing between professional developers and hobbyists may become crucial in mitigating unintended consequences from misinterpretation of intent.