How much more software do we really need?

from Aarva · host Noah Smith

With AI creating so much code, where are all the new products?We’re all hearing about AI’s incredible productivity gains, especially in coding. But what if all that “tokenmaxxing” isn’t actually translating into new, useful products? This piece dives into the curious disconnect between the massive investment in AI coding agents and the surprisingly quiet impact on what we actually *see* and *use*. It makes you wonder if we’re just spinning our wheels, or if AI’s true revolution in software lies elsewhere entirely—a question that has big implications for the future of work and innovation.An analysis of the rapid adoption and significant investment in AI coding agents observes a current disconnect between high token usage and the actual delivery of new software products. It explores the challenges of translating task-level AI productivity into economic value, suggesting this phenomenon is partly normal experimentation. The piece concludes by considering whether AI's major impact will manifest in entirely new software industries rather than merely improving existing ones.Read at source: Noahpinion

What this episode covers

With AI creating so much code, where are all the new products? We’re all hearing about AI’s incredible productivity gains, especially in coding. But what if all that “tokenmaxxing” isn’t actually translating into new, useful products? This piece dives into the curious disconnect between the massive investment in AI coding agents and the surprisingly quiet impact on what we actually *see* and *use*. It makes you wonder if we’re just spinning our wheels, or if AI’s true revolution in software lies elsewhere entirely—a question that has big implications for the future of work and innovation. An analysis of the rapid adoption and significant investment in AI coding agents observes a current disconnect between high token usage and the actual delivery of new software products. It explores the challenges of translating task-level AI productivity into economic value, suggesting this phenomenon is partly normal experimentation. The piece concludes by considering whether AI's major impact will manifest in entirely new software industries rather than merely improving existing ones. Read at source: Noahpinion