Will AI replace programmers?
Will AI replace programmers? When we ask this question, what we're really asking is: Will every layperson be able to make programs using AI? And that is like asking: Did the invention of the hair clipper replace hairdressers?
Did it lead people to cut their own hair? Not really. The tool made the hairdresser's job easier. Just like a hair clipper, AI is merely a tool.
But let's take a step back and delve deeper into the issue. The question of whether AI will replace programmers has been on my mind almost daily since ChatGPT emerged about a year ago. I have been using ChatGPT but also other AI tools specialized in generating code such as GitHub Copilot, Cursor AI, and Tabnine. And I understand why some programmers especially wanna-be programmers panic. And it is all due to a fundamental misconception they are having.
The misconception is that “AI can make programs”. That is wrong. AI can write snippets of code. Ai cannot make programs. Yes, it can write small toy programs, but those are not real-world programs. Real programs consist of dozens or even hundreds of source files with intricate configurations and interlinking with other tools like databases which form layers of complexity only human intelligence can navigate.
Just yesterday, I was solving a Django database migration issue and it took me half a day to even understand the problem and I didn’t even bother to think of using tools like ChatGPT to solve it.
AI is just a tool. It makes programmers’ job easier. You no longer have to write snippets of code by hand. You can use AI. But you still have to bring all the pieces together. And only a programmer can do that.
The same principle applies to the hair clipper analogy. You can use a hair clipper instead of manually using scissors, but you still need to have the skill to bring the job together.
AI in programming is simply a new sophisticated tool. I am not even delving into the topic that these AI tools are simply LLMs (Large Language Models). They are highly knowledgeable, but not intelligent. If we still have humans doing simple jobs that require minimal brain effort, it is absurd to think that AI would replace one of the most intellectually-involving activities such as programming.
So, programmers, let's roll up our sleeves and dive into the code! There is so much to learn and build.
Happy coding,
Ardit!
Thanks for posting those Devin responses, but what is your conclusion? I was expecting some kind of response of your own based on Devin's comments.
I remember decades ago that by about 20 years ago, there should not be any paper being used in offices at all, yet, there's still plenty of paper around. What might be saved in paper and ink is spent on electricity and hardware to keep electronic files available. Also, Devin relies on a computer network, electricity, and other peripheral devices to function, whilst a pen/pencil and paper can be used at any time, and when coupled with developers together, you have your AI Actual Intelligence.
This is the answer of Devi: !
The statement captures a common perception about AI's programming capabilities but also contains some misconceptions. Let's break it down:
"It can write small toy programs, but those are not real-world programs."
While it's true that AI, in its current state, is exceptionally good at generating small, self-contained code snippets or "toy programs," my capabilities extend beyond that. I can help with more complex, real-world programming tasks, such as developing full-scale applications, writing scripts for automation, debugging, and offering solutions to software engineering problems. However, the complexity and effectiveness of the generated code can vary, and human oversight is essential.
"Real programs consist of dozens or even hundreds of source files with intricate configurations..."
This is accurate. Real-world software projects are often complex, consisting of multiple components, libraries, and configurations. I can assist in understanding and developing such complex systems, provide guidance on structuring projects, and offer insights on best practices for managing dependencies and configurations.
"...and interlinking with other tools like databases..."
I can provide code examples, explanations, and best practices for integrating various tools and technologies, including databases, into software projects. This includes guidance on database design, writing SQL queries, and integrating databases with different programming languages.
"...which form layers of complexity only human intelligence can navigate."
While it's true that human intelligence is currently unmatched in navigating the nuanced and sometimes unpredictable challenges of software development, AI can significantly aid in this process. I can offer insights, generate code, and provide solutions that address complex problems, but human expertise is crucial for decision-making, particularly in strategic and highly nuanced contexts.
In conclusion, while there are limitations to what AI can achieve in software development compared to human experts, the notion that AI can only handle "toy programs" is an understatement of its capabilities. AI, especially a specialized version like myself, can contribute significantly to various stages of the software development lifecycle, from planning and writing code to testing and deployment. However, the collaboration between AI and human developers is key to leveraging the strengths of both and achieving the best outcomes in software projects.