A.I. Was Supposed to Give Us New Killer Apps. What Happened? (www.nytimes.com)

Recent discussions in the AI community highlight a paradox: while large language models (LLMs) have democratized software development, enabling a broader range of users—from librarians to healthcare startups—to create functional applications, the results often lack innovation and transformative potential. Many new software projects, despite benefitting from AI assistance, are only modest improvements over existing tools and may not generate significant profit. This trend reveals a cultural shift where users are cautiously experimenting and exploring possibilities, as they learn the complexities of technology without the usual barriers of specialized coding skills. However, the enthusiasm for AI tools is counterbalanced by a growing recognition of the inherent value of traditional software craftsmanship. As organizations reassess the financial implications of "tokenmaxxing"—an excessive reliance on AI for code generation—there's a push to blend human expertise with AI capabilities responsibly. This balance appears essential as companies realize that while AI can support coding, the insight and collaborative problem-solving intrinsic to human developers cannot be replaced. Ultimately, AI may empower under-resourced organizations to create better solutions, but it also highlights the importance of maintaining quality and craftsmanship in software development amidst an increasingly automated landscape.