One of the biggest breakthroughs in the field of AI this year should be in the field of programming, AI programming tools like Cursor and v0 dev have not only drastically lowered the threshold of programming for ordinary people, but also allowed professional programmers to dramatically increase their development efficiency.
But we hear the news are not programming high school students, product managers, with the help of AI programming tools in a few hours to make a hot product, but did not hear that there are programmers because of the programming efficiency and promotion and salary increase, but on the contrary, there are more for the AI will replace the programmers of the worry.
This is because AI programming, while enhancing development efficiency, dilutes the scarcity of writing programs and begins to change the paradigm of software development so that the single chain that goes from requirement to product begins to fork into multiple branches, with most requirements being solved in the first half of the chain without having to be passed on to the second half, or even without the need for professional programmers to be involved.
What does the traditional requirements development model look like?
The traditional requirements development model is like a chain: Requirements -> Product Design -> Development -> Testing -> O&M.
When the user has a need, such as to translate an article, to deal with the merger of PDF files, which require professional software support, the development of such professional software, there is a need for professional product managers will be the original needs of the user, the design can be used for the user to facilitate the operation of the UI interface, programmers and then based on the product manager's design, to do the design of the system to write the code, and finally made to meet the needs of the software.
In this chain, product managers and programmers are both highly specialized positions, extremely difficult to replace each other, even for very simple apps, ordinary product managers can not play the role of programmers, but in turn, ordinary programmers can not do the work of product design. Of course, there are very few "independent developers" who can do both.
The changes AI brings to the requirements development paradigm
Simple requirements don't require software development anymore
The first change is that simple needs no longer require "software" and can be solved by "chatting" directly with the AI.
We used to need specialized software to translate, but now we send the text to the ChatGPT or Claude Such AI tools, very quickly translated for us; or a paragraph of the report needs to be inside the structured data extraction, before to write scripts or tools, is now also sent to the AI on it; such examples are many, some of the common small needs of our daily life, it is very easy to use the AI chat tool to meet.
And AI models are also being upgraded, before ChatGPT, Claude they can only chat, now you can upload pictures, documents, generate documents, and in the future can also generate videos, but also can execute code, virtual machine to run a Python script, or generate a beautiful report, can do more and more things.
Ordinary requirements can be initiated without relying on professional programmers
The second change is that with product design, you can start a project and make a prototype even if you don't have a programmer.
In the past, a popular terrain is: "everything is ready, just missing a programmer", after all, just product design is not enough, you need a programmer to realize the product design, but now with AI assistance, even without programming foundation, you can make a simple application, or prototype.
Many programmers do not look at these AI-developed products, that is not professional enough, just prototype products, but they can really meet the needs of many users are not so picky, to meet the needs of the good, maybe not stable maybe not good-looking, but can solve the problem.
Recently very hot kitten fill light, the important thing is not that it is made with AI, not the author's occupation is a product manager is not a professional programmer, but it can meet the needs of users, users are willing to pay money. As for the use of AI, the essence is just a tool to realize the demand and marketing.
There will be more and more such cases in the future, and there will even be an explosion of small and beautiful applications, because the decomposition and production of requirements will no longer require the participation of professional programmers at the initial stage, and non-professionals will be able to convert ordinary requirements into prototypes. Once the requirements are proven to be feasible, either the author can further learn professional programming skills or find programmers to work with.
Why are so few programmers successful as indie developers these days? Because programmers are too far away from the needs, too far away from the users, and don't really know what the users want or how to sell it to them!
In fact, the user has a lot of needs pain points have not been met, but before the bitter no technology can be realized, and from now on, many of the needs of the user can with the help of AI, their own needs into a tool, first to meet their own needs and the needs of people around them, part of which will be fire out of the circle to go.
Complex requirements will still need to be designed by professional programmers, but the development process will be made much more efficient by AI.
The third change is that AI will dramatically improve the development efficiency of professional programmers, which in turn will shorten software project development cycles.
Software engineering has gone through several major changes so far:
- Waterfall model: truly marks the beginning of software engineering, with a clear division of labor, a software project lifecycle, making software development measurable
Agile development: allowing software projects to go small, allowing for rapid iteration and rapid delivery, and better response to changes in requirements
DevOps: continuous delivery, continuous integration, so that the entire development, integration and deployment of automation, development, testing and operation and maintenance from the heavy manual work to free, from the demand side to the final release of the automation of the entire process, and through automated testing to ensure the quality of the software
Next AI will trigger another change in software engineering, making software development intelligent. This intelligence process will be divided into several stages.
The first phase is the one we're going through, where AI programming tools are helping programmers dramatically improve development efficiency. Last year GitHub Copilot It gives me the impression that it will approximately improve the development efficiency of the 10%-20% this year Cursor Already I've been able to have 30%-50% efficiency gains, depending of course on the type of project and the proficiency of the user, but the trend is clear.
The second stage is coming soon, is that AI is not only in the field of programming, in other areas such as testing and operation and maintenance will also greatly improve efficiency, a lot of automated test code will be generated by AI, Claude's computer use of such technology maturity, a lot of previous only manual testing work will be able to be completed by AI, manual as long as a small amount of checking can be.
AI can also play an important role in log analysis and fault recovery troubleshooting for online operations and maintenance in the future.
The third stage is that in the future there will be a new software architecture and programming language for AI technology. Traditional software architecture and programming language are designed for human beings, and now AI is trying to accommodate human beings and program in a human way, which is not necessarily the most suitable way for AI.
The current generative AI just generates text, images and videos, and the future AI should be able to directly and dynamically generate UI interfaces and game screens, which will also allow programming to be further naturalized, and more complex software and games can be made through natural language.
What the paradigm shift in requirements development has taught us
Every such change is a challenge and an opportunity.
For ordinary people, some pain points can be solved with the help of AI, not only with the help of AI chat tools, but also with the help of AI programming tools to write some small scripts and gadgets to directly solve the pain points, significantly improve their work efficiency, and even create a hot product.
For product managers, instead of having to stop or be limited to the product design field, they can go further and make usable prototypes of their products with the help of AI programming tools to quickly validate requirements.
For programmers, the future mastery of AI programming tools to enhance efficiency is essential, otherwise there is a risk of elimination, actually use it easier than imagined may be, are not required to spend money to buy courses, as long as the psychological does not resist more use more experience can be.
On the other hand, if the programmer can go more contact with the user, more to find around the demand, do not need to go to the roll notes, bookkeeping, ToDo three-piece suite of such a demand for bad street products, with the help of AI can be quickly realized to deliver, I guarantee that you do out of the result is certainly better than non-professional programmers a lot.
Although the employment situation is not good now, but standing in the global market, in fact, the demand for games, apps, websites are very large, good discovery can find a lot of opportunities, first go to find the demand, and then use AI to quickly realize the on-line, rapid trial and error to accumulate experience, you can certainly find their own opportunities.
Regardless of your profession, if you want to seize opportunities in the midst of change, the most important thing is to keep learning and adapting, to understand the latest AI tools and capability boundaries, to master the best practices of collaborating with AI, and to enhance your cross-border capabilities with the help of AI without being limited to the profession you are engaged in.
Regardless of how the requirements development paradigm of software engineering changes in the future, the nature of value remains the same - the essence of creating value is still helping users solve real problems. the advent of AI tools allows us to validate ideas faster and solve problems more efficiently, but ultimate success still depends on creating real value for users.
The paradigm shift brought about by AI is not eliminating certain types of roles, but reshaping the entire software development ecosystem. Future success belongs to those who can understand this shift and are adept at utilizing the new tools to create value.
