The opinions included are those of Daniel Stenberg, the creator and principal maintainer of cURL, as well as the creators of Python, Svelte, and other widely used open-source projects.

Among the researchers quoted is Johannes Wachs, Research Fellow at ANETI Lab and Associate Professor at Corvinus University of Budapest, whose work informs the article’s discussion of programming skills and public knowledge-sharing. The article highlights two closely connected challenges.

First, generative AI has made it much easier to produce code and submit proposed changes to open-source projects. Reviewing, correcting, and maintaining those contributions, however, still requires skilled human work. Maintainers are increasingly faced with AI-generated submissions that may be inaccurate, unnecessary or poorly adapted to the wider context of a project. Instead of reducing their workload, these contributions can transfer additional costs to the people responsible for keeping essential software functioning.

Second, AI coding tools raise questions about programming skills and the future of publicly shared technical knowledge. Although they can improve productivity, particularly among experienced programmers, users still need sufficient expertise to evaluate the code they generate.

As Johannes explains:

“To be able to verify the outputs of a large language model, you need to understand something about the code.”

This point relates to the study Who is using AI to code? Global diffusion and impact of generative AI, co-authored with Simone Daniotti, Xiangnan Feng and Frank Neffke and published in Science. The study finds that experienced programmers capture most of the productivity and exploration gains associated with AI-assisted coding, potentially widening rather than closing existing skill gaps.

The FT article also refers to Johannes’s research with R. Maria del Rio-Chanona and Nadzeya Laurentsyeva on the decline of public knowledge-sharing after the introduction of ChatGPT. Their study, published in PNAS Nexus, documents a substantial fall in activity on Stack Overflow. Johannes notes in the feature that programming-related human interaction has not disappeared, but increasingly takes place through private AI platforms rather than contributing to publicly accessible knowledge.

Another study discussed in the article is Vibe Coding Kills Open Source, co-authored by KRTK colleagues Miklós Koren and Gábor Békés, together with Julian Hinz and Aaron Lohmann. The paper examines how AI agents may weaken the incentives that sustain open-source development: software packages can gain machine-generated downloads without producing human engagement, recognition or financial support on which many maintainers rely.

Together, these contributions point to a central tension: AI is making code easier to produce, while maintaining software, developing the expertise needed to evaluate it and sustaining shared knowledge resources remain fundamentally human and collective tasks.

The original Financial Times feature is available to subscribers.

Related publications

Who is using AI to code? Global diffusion and impact of generative AI
Simone Daniotti, Johannes Wachs, Xiangnan Feng and Frank Neffke, Science, 2026

Large language models reduce public knowledge sharing on online Q&A platforms
R. Maria del Rio-Chanona, Nadzeya Laurentsyeva and Johannes Wachs, PNAS Nexus, 2024

Vibe Coding Kills Open Source
Miklós Koren, Gábor Békés, Julian Hinz and Aaron Lohmann, 2026