We (I) need science
Argues for rigorous scientific studies on the impact and risks of using LLMs in software development, highlighting current lack of impartial research.
Tony Arnold is an Australian software developer based in Newcastle, working at Mantel on Reveal, a visual debugging tool for mobile applications. He writes about software development, mobile experiences, and open-source, and actively contributes to community projects.
4 articles from this blog
Argues for rigorous scientific studies on the impact and risks of using LLMs in software development, highlighting current lack of impartial research.
Analyzes the hidden costs and skill erosion of using AI for coding, emphasizing the need for human oversight.
A critique of relying on others' opinions about coding assistants, urging developers to trust their own skills and curiosity.
Introduces Swift Dependency Audit, a CLI tool and SwiftPM plugin for auditing Swift package dependencies against source imports.