Lawson argues AI coding tools can yield high-quality code with deliberate pacing
New opinion piece suggests large language models offer flexibility for superior output, though the process requires more time than typical automated workflows.
Technology writer Nolan Lawson has published a perspective challenging the dominant view within the software development community regarding the utility of artificial intelligence coding assistants. In an article posted on his personal website, nolanlawson.com, Lawson contends that large language models (LLMs) are not exclusively tools for generating substandard code at speed.
The prevailing sentiment among many developers is that AI coding software is primarily designed to produce low-quality output rapidly. This approach often involves generating barely functional code, opening extensive pull requests, and merging them without adequate vetting. Lawson’s piece, titled “Using AI to write better code more slowly,” directly counters this narrative by highlighting the versatility of current AI technologies.
Lawson argues that LLMs possess significant flexibility, allowing them to be utilised effectively for the creation of high-quality code. However, he notes that achieving this standard requires a different workflow than the rapid generation often associated with these tools. The process of writing better code using AI is inherently slower, demanding a more deliberate and careful approach from the developer.
The article was published on 25 May 2026 and has drawn attention through community platforms such as Hacker News. It positions itself as an opinion piece rather than empirical data, offering a subjective observation on how AI tools can be integrated into software development practices to prioritise quality over speed.
This perspective adds nuance to the ongoing discussion about the role of AI in finance and technology sectors, where code integrity is paramount. By suggesting that developers can leverage LLMs for superior outcomes, Lawson provides a counterpoint to the fear that automation inevitably leads to a decline in code standards.
