Can GPT Shrink a Ten‑Person Development Team to a Few in the Next Three to Five Years?

In a recent Zhihu question asking whether GPT could reduce a ten‑person development team to just a few within three to five years, the author shares personal reflections on the impact of ChatGPT on programming.

ChatGPT excels at design and code generation, allowing developers to guide it to produce module‑level code, but developers must still verify and validate the output. The author emphasizes that, with ChatGPT assistance, programmer productivity can increase dramatically—code generation can be ten times faster, and even after accounting for verification and debugging, a three‑to‑five‑fold boost is realistic.

Adoption of ChatGPT currently has a barrier: early adopters reap significant efficiency gains, while many others remain skeptical or indifferent. Over time, most developers are expected to join the "AI‑generated code" camp because not using the tool puts them at a competitive disadvantage in the eyes of employers.

The author argues that, unlike past technological revolutions (e.g., the introduction of Fortran), which expanded the overall demand for programmers rather than shrinking the workforce, the total demand for IT talent continues to grow. Historical shifts—from mainframes to PCs, from PCs to the internet, and now to mobile and cloud—have each opened new markets that required more developers.

If large‑scale models like ChatGPT eventually become general artificial intelligence, they could theoretically replace programmers, affecting all industries. However, at present, ChatGPT is a productivity tool that will not directly shrink the professional developer community. The key factor will be how AI models create new opportunities across various sectors.

The author remains optimistic, likening the rise of ChatGPT to the transformative impact of the 1990s internet boom, expecting it to trigger profound technological change. Specialized AI models (e.g., MathGPT, medical GPTs) will emerge, leading to new roles such as developers who train industry‑specific models and those who build applications on top of them.

Consequently, while some programming jobs may disappear, new positions will arise, and the market will continue to expand. The author warns that without new technological breakthroughs, the programming community could face stagnation, layoffs, and intense competition.

Finally, the article includes a promotional notice for a Thursday evening live session featuring Microsoft senior Data & AI cloud solution architect Zhao Mingjie discussing "GPT Application Innovation Trends," with a QR code for joining the event.