Warning: GitHub CEO issues an "ultimatum": "Either embrace AI or get out of the industry."

This week, a statement by GitHub CEO Thomas Dohmke on the social platform X (formerly Twitter) sparked a heated discussion among developers:

“The evidence is crystal clear: either embrace AI or exit the industry.”

This statement is not Thomas Dohmke scaremongering. Instead, it is a conclusion he reached based on a recent in - depth study by GitHub on users of AI tools. As he said in his latest blog post, “Developers, Reinvented,” AI is reshaping the roles, skills, and working methods of developers at an astonishing speed.

01

No Longer Just “Code Writers,” but “AI Managers”

In the past, programmers would always immerse themselves in editors, writing code, debugging, and testing. However, nowadays, AI tools have quietly intervened in the workflow. From auto - completion to code generation, from debugging assistants to project partners, more and more developers are starting to view AI as a “collaborator” in their work.

GitHub's study interviewed 22 developers who use AI tools extensively in their daily work.

According to the study, GitHub divides the process of developers adopting AI into the following four stages:

(1) AI Skeptics: They try AI for the first time, using it to complete a few lines of code or write a function. However, they have a low tolerance for errors and are likely to give up halfway.

(2) AI Explorers: They use AI to write boilerplate code, debug, and refactor, gradually understanding its capabilities and limitations.

(3) AI Collaborators: They jointly develop with AI and learn to use context engineering to improve AI's understanding.

(4) AI Strategists: They regard AI as the core productivity in development, using it to assist in completing complex functions, large - scale refactoring, and system design.

Among them, developers who have entered “Stage Four” unanimously believe that “the role of developers has undergone a fundamental transformation.”

Put simply, they believe they have entered a brand - new professional stage: no longer just code producers, but system builders, AI Agent schedulers, and result reviewers. Their daily work now consists of two core aspects:

● Delegation: Provide context for AI, design prompts, and set task goals to guide agents to produce results in the expected direction.

● Verification: Critically check whether the code generated by AI meets the specifications, whether the logic is correct, and whether the boundary conditions are clearly considered.

Obviously, AI is no longer just an “auxiliary tool,” and even its “role division” has undergone a fundamental change. As one developer said, “Maybe in the future, we won't be code producers anymore, but code 'empowerers'. My next title might be 'Code Creative Director'.”

02

How Soon Will the Era of “90% of Code Generated by AI” Arrive?

In this survey, GitHub also asked developers a question: “When do you think the era of '90% of code generated by AI' will arrive?” The result was unexpected: half of the developers believe it will be achieved within 5 years, while the other half think it will only take 2 years.

The most crucial point is that they are not threatened by AI. Instead, they are both excited and looking forward to it.

“These developers don't feel that their value has been diminished. Instead, it has been'redefined,'” Thomas Dohmke added on X. “Developers are not anxious. They are both realistic and optimistic about the future.”

Interestingly, the GitHub study also found that although many people assume that “the advantage of AI lies in saving time,” in fact, developers value the “capability leap” brought by AI more. As Thomas Dohmke summarized, “With the support of AI, time is indeed saved. But more importantly, developers are constantly 'raising the ceiling,' not just blindly 'cutting costs'.”

This also explains why many developers are willing to “pay for” the top - tier packages of AI tools - not for being lazy, but for challenging greater goals.

Based on the above research results, Thomas Dohmke summarized seven core capabilities that developers must master in the AI era:

(1) AI Literacy: Understand different AI tools and models;

(2) Agent Collaboration Ability: Set up the execution environment and prompts for AI;

(3) Human - Machine Collaboration: Be able to collaborate with AI in real - time;

(4) Basic Programming Skills: Master solid programming design capabilities;

(5) Verification and Quality Control Ability: Carefully check the work results of AI;

(6) Product Understanding: Not only understand code but also have systematic thinking;

(7) Architecture Design Ability: Know how to integrate AI into the entire system structure.

He also specifically pointed out that “a vague prompt cannot make AI produce high - quality results, just as you can't give a vague instruction to a colleague and expect them to complete the task perfectly.” At the same time, if beginners only rely on prompts but don't understand the essence of code, they will eventually hit a ceiling. By then, “you need to truly understand the code, debug it, and make it scalable.”

In addition to individual developers, Thomas Dohmke also called for the reconstruction of the computer science education system.

“Students will rely on AI to complete more and more code tasks. At this time, teaching by examining grammar memory and API proficiency is already outdated,” he said bluntly on X. “Just teaching grammar is no longer enough. Now students must learn how to guide AI, evaluate AI's output, and think interdisciplinary.”

“It's not about 'writing this loop,' but about 'understanding what this code does and which parts will be affected if it is changed.'”

03

The Software Industry Won't Be Eliminated by AI but Is Being Reborn

Although there are widespread claims that “AI will replace programmers,” real - world data shows another possibility: The U.S. Bureau of Labor Statistics predicts that software development positions will increase by 18% in the next decade, nearly five times the average growth rate of the entire industry.

Thomas Dohmke agrees with this data, but he also admits that “these won't be the developer positions we are familiar with.”

As early as July this year, he said, “Smart companies will hire more developers because if one person can achieve 10 times the efficiency through AI, then 10 people may achieve 100 times the output.” And this time, Thomas Dohmke also added that those hired must be developers who have adapted to the AI transformation.

Of course, Thomas Dohmke also admits that not every developer is willing to embrace this change: “For some people, managing AI agents to complete tasks may not sound very rewarding.”

But he pointed out that in essence, programmers have always been using languages to communicate with computers. The only difference now is that this “language” has become a prompt, and the “communication object” has become AI. Thomas Dohmke summarized, “This change is not a hypothesis but a reality. We see developers moving from doubt to trust and reshaping their work methods and mindsets.”

If we follow Thomas Dohmke's statement, then for developers who still doubt AI, this is both a warning and a roadmap: the transformation is happening, and the earlier they adapt to AI, the more advantageous their position will be in the future software industry.

04

Netizens Question: Can a Study Involving 22 People Represent the Entire Industry?

However, Thomas Dohmke's remarks have also raised doubts and resentment among some developers:

“This 'either embrace artificial intelligence or give up' ultimatum is a typical false dilemma - you are ignoring the entire middle ground where most successful developers are. A study involving only 22 people cannot support a judgment on the future of the entire industry.”

“Don't be so black - and - white. What we should really refer to are those developers who can integrate AI tools thoughtfully while maintaining their core programming skills. Their performance will definitely be better than those who 'fully accept AI' and those who firmly resist AI.”

However, more developers highly agree with Thomas Dohmke's analysis and conclusions:

“In my opinion, developers really don't need to worry. AI should be regarded as an accelerator that can increase your skills and output by 10 times, not a replacement. For example, I can now complete tasks that I couldn't do before at an unprecedented speed... This is very exciting, and everyone should feel the same way.”

“Indeed, I'm completely different with AI compared to without it.”

“I completely agree with your view. AI has opened the door to numerous tools and technologies for me and helped me enter fields that would have required a lot of time and effort to access.”

So, what's your opinion on this?

Reference Link:

https://www.finalroundai.com/blog/github - ceo - thomas - dohmke - warns - developers - embrace - ai - or - quit

This article is from the WeChat official account “CSDN”, author: Zheng Liyuan. It is published by 36Kr with authorization.