Tesla CEO Elon Musk has highlighted the strategic importance of technology in modern warfare, responding to veteran programmer John Carmack’s argument that artificial intelligence could fundamentally change the role of traditional coding. As AI systems take on increasingly complex programming tasks, the debate is shifting from whether developers will use AI to how much traditional software engineering expertise will remain essential.
Musk: Technology Is the Critical Advantage in Modern Warfare
Tesla CEO Elon Musk entered the discussion after John Carmack, the veteran programmer and founder of AI startup Keen Technologies, published an extended argument about the changing nature of software development.
Musk responded with a concise observation:
“Technology is the most important advantage in war.”
The comment comes as artificial intelligence becomes increasingly intertwined with military research, autonomous systems, communications, intelligence analysis and software development. While Musk did not specifically outline a military AI strategy in his response, his statement reflects a broader argument that technological superiority can increasingly influence the outcome of conflicts.
At the same time, Carmack’s original post focused primarily on programming rather than warfare. His argument was that AI is reducing the importance of manually writing software line by line, potentially transforming programming from a core professional skill into a specialized discipline.
John Carmack Compares Manual Coding to Martial Arts
Carmack, whose career includes decades of work in computer programming and video game development, used an unusual martial arts analogy to explain how he believes programming is changing.
His argument centers on the distinction between martial arts practiced primarily for combat and martial arts practiced as a discipline. In Japanese terminology, the distinction between “jitsu” and “do” is often used to describe practical technique versus a broader path of personal development.
Carmack suggested that AI could push programming in a similar direction.
In his view, programmers may increasingly rely on AI systems to generate, modify, test and debug software instead of manually producing every line of code themselves. The ability to understand programming would remain valuable, but the act of typing code could become less central to professional software development.
He also warned programmers against becoming overly attached to traditional methods simply because those methods worked in the past.
His broader point is that technological changes do not necessarily eliminate expertise. Instead, they can change which parts of that expertise provide the greatest value.
A programmer who once spent most of the day writing code might increasingly spend time defining system requirements, reviewing AI-generated implementations, testing outputs, identifying security weaknesses and deciding how different software components should interact.
AI Is Moving From Coding Assistant to Development Partner
The shift Carmack describes is already visible across the technology industry.
Generative AI coding systems can produce functions, write tests, explain unfamiliar code, identify potential bugs and convert natural-language instructions into software. As these systems become more capable, developers can potentially complete tasks that previously required substantially more manual work.
This does not necessarily mean that programming knowledge has become irrelevant.
AI-generated software still needs to be evaluated. A model can produce code that appears correct while containing subtle logical errors, security vulnerabilities, inefficient implementations or incorrect assumptions about the underlying system.
That creates an important distinction between writing code and engineering software.
Writing code is only one component of software engineering. Developers also need to understand architecture, data structures, security, performance, reliability, testing and the requirements of the people using the system.
As AI becomes better at producing code, those higher-level skills could become more important rather than disappearing entirely.
Anthropic Predicts a Rapid Transformation of Software Engineering
Carmack’s comments follow increasingly aggressive predictions from executives at major AI companies.
Anthropic CEO Dario Amodei has argued that AI could handle a substantial portion of software engineering work, while describing his own reduced reliance on manually written code as evidence of how quickly development practices are changing.
The underlying prediction is significant because software engineering has historically been considered a highly specialized profession requiring years of training.
If increasingly capable AI systems can perform much of the implementation work, companies could potentially reorganize development teams around smaller groups of highly skilled engineers directing AI systems.
However, forecasts about the speed of this transformation remain predictions rather than established outcomes. The complexity of production software, security requirements, regulatory obligations and the need for human accountability can make real-world development considerably more difficult than generating code in a controlled demonstration.
OpenAI and Google Show How Quickly AI Coding Is Expanding
Other technology executives have reported similarly rapid increases in AI-generated code.
OpenAI President Greg Brockman has described a dramatic increase in the amount of code produced with AI assistance. Alphabet and Google CEO Sundar Pichai has likewise said that a substantial portion of new code at Google is now generated with the assistance of AI.
These developments point toward a changing division of labor.
Instead of developers acting as the sole producers of software, AI systems can increasingly function as collaborators that translate specifications into implementations. Developers then supervise the process, evaluate the output and make decisions that require broader technical judgment.
That model could eventually make the ability to communicate precise instructions to AI systems nearly as important as traditional programming syntax.
Not Everyone Believes AI-Assisted Coding Is an Improvement
The transformation has also attracted significant criticism.
Minecraft creator Markus Persson has criticized AI-assisted programming, arguing that the technology can encourage developers to focus on producing code rather than understanding the underlying logic.
That criticism highlights one of the biggest unresolved questions surrounding AI-powered development.
If inexperienced programmers can generate sophisticated applications without understanding how the underlying systems work, they may be able to build products quickly but struggle to diagnose failures when something goes wrong.
There is also a potential talent-development problem. Junior developers traditionally learn by writing relatively simple programs, debugging mistakes and gradually taking responsibility for increasingly complex systems.
If AI performs much of that entry-level work, companies may need to find new ways to train the next generation of engineers.
The Value of Programming Knowledge Could Change
The emerging AI model does not necessarily point toward the disappearance of programmers. It could instead redefine what makes a programmer valuable.
Traditional skills such as syntax memorization and manually implementing routine functions may become less important as AI takes over repetitive tasks.
Meanwhile, other abilities could become more valuable:
- System architecture: Designing how complex applications and services work together.
- Code review: Determining whether AI-generated software actually performs as intended.
- Security: Identifying vulnerabilities that automated systems may overlook.
- Debugging: Understanding why a system fails when AI-generated solutions do not work.
- Product judgment: Translating real-world requirements into technical specifications.
- AI supervision: Directing multiple AI systems and evaluating their outputs.
- Deep technical knowledge: Understanding the principles necessary to recognize when an AI-generated answer is fundamentally wrong.
In that environment, the best engineers may not necessarily be those who type the fastest. They may be those who can determine what needs to be built, why it should be built that way and whether the resulting system can actually be trusted.
Technology, AI and the Future of Warfare
Musk’s brief comment about technology being the most important advantage in war adds another dimension to the discussion.
Modern military competition increasingly involves software, autonomous systems, satellite networks, cybersecurity, intelligence processing and artificial intelligence. The ability to develop and deploy advanced technology can therefore have consequences far beyond conventional software development.
AI-assisted programming could accelerate the creation of these technologies by allowing smaller teams to develop increasingly sophisticated software.
That possibility also creates risks. Faster software development can accelerate beneficial innovation, but it can also reduce the time available to identify vulnerabilities, assess unintended consequences or establish appropriate safeguards.
The same underlying technology that allows an engineer to build an application faster could potentially be used to develop systems with far greater consequences.
The Bigger Shift: From Writing Software to Directing Machines
Carmack’s argument ultimately points toward a broader transformation in the relationship between humans and computers.
For decades, programmers translated human ideas into machine-readable instructions. AI is beginning to reverse part of that relationship. Humans can increasingly describe an objective in natural language while an AI system translates that objective into executable software.
That does not eliminate the need for human expertise. Instead, it changes where that expertise is applied.
The emerging question is therefore not simply whether AI will replace programmers. It is whether programming itself will become less about manually writing instructions and more about directing, testing and governing intelligent systems that write those instructions on a developer’s behalf.
Musk’s observation about technology and Carmack’s warning about traditional coding ultimately intersect at the same point: the competitive advantage may increasingly belong to people and organizations that learn how to work effectively with increasingly capable machines.
