How I Use "AI" (2024) (nicholas.carlini.com)

In a reflective blog post, a programmer and research scientist shares how interactions with large language models (LLMs) have significantly enhanced his productivity, challenging the contrasting narratives of over-enthusiasm and skepticism surrounding AI. While he acknowledges that the term "AI" is often misused and implies a level of intelligence that LLMs do not possess, he asserts their practical utility in his work. Over the past year, he has utilized LLMs to automate mundane coding tasks, learn new frameworks, and enhance his research output, claiming a productivity boost of roughly 50%. This revelation counters prevalent pessimism about LLMs' contributions to the field. The significance of this account lies in its potential to shift the AI/ML conversation towards nuanced, practical applications of LLMs rather than apocalyptic predictions or claims of impending job losses. The author emphasizes that LLMs have revolutionized his approach to coding by enabling him to build complex web applications and streamline his workflow, underscoring their role as tools for learning and productivity. By detailing specific examples of his interactions with LLMs, he provides a grounded perspective that encourages professional communities to consider the concrete benefits of embracing these technologies while remaining critical of their limitations and ethical implications.