To be fair, it had already started happening much before AI came when programmer roles started getting commoditized into “Python coder”, “PHP scripter”, “dotnet developer”, etc. Though these exact phrases weren’t used in job descriptions, this is how recruiters and clients started referring programmers as such.
But LLMs took it a notch even further, coders have started morphing into LLM prompters today, that is primarily how software is getting produced. They still must baby sit these LLMs presently, reviewing and testing the code thoroughly before pushing it to the repo for CI/CD. A few more years and even that may not be needed as the more enhanced LLM capabilities like “reasoning”, “context determination”, “illumination”, etc. (maybe even “engineering”!) would have become part of gpt-9 or whatever hottest flavor of LLM be at that time.
The problem is that even though the end result would be a very robust running program that reeks of creativity, there won’t be any human creativity in that. The phrase dismal science was first used in reference to economics by medieval scholars like Thomas Carlyle. We can only guess their motivations for using that term but maybe people of that time thought that economics was somehow taking away the life force from society of humans, much similar to the way many feel about AI/LLM today?
Now I understand the need for putting food on the table. To survive this cut throat IT job market, we must adapt to changing trends and technologies and that includes getting skilled with LLM. Nonetheless, I can’t help but get a very dismal feeling about this new way of software development, don’t you?
I think we are talking about 2 different things. Coding an entire app or deliver code snippets only.
Coding an entire app? I have tested AI coding for a small app and seen the result. Thousands lines of code that I never could maintain. My feelings is the same as asking a three years old child to drive my car - similar to hand over the app coding to AI. I have no fear that this will replace me or many others in the near future. (What I have seen so far)
Code Snippets? Yes, they are good code buddies and are really helpful. Will this replace any coders? Of course. You as developer can code faster and deliver more will reduce the need for new colleagues.
This would be true if, and only if, you actually let the AI do all the work for you and dont actually USE the code it generates yourself.
If you start getting into the habit of sending the requirements in, getting the output, and blindly using it, you will be one of those things. And what you output will be worse, because you wont be able to maintain it.
It has started happening in many companies. For example, the Anthropic CEO recently said that AI will be writing 90% of their code in a matter of just 3-4 months. Similarly, the Github CEO has also said that 80% of their code will be written by AI.
Yes the management would love to see the AI replacing expansive employees. But it’s the same what happens to one of my projects now. It has moved from EMEA to International now. I maintained and developed the tool for over 10 years now and I am one of maximum 3 in the whole company who knows the workflow processes of the daily work it is used for behind this. International now means it’s a good idea to save money and no longer hire me but replace me with an Indian team of one scrum master, one data analyst and two programmers, which have no idea what the software is good for, or what the enduser is doing with it.
I have been programming for more than half a century. I know that computer software has gotten really complicated.
This post sounds more like a rant than a real question. There are multiple issues.
People are constantly creating new words and phrases for old things.
First, the term LLM is a term that is morphing. I asked AI questions related to the definition of AI and it used the term LLM in a more general manner than I thought it was and I said I was confused and it admitted it was not using the term properly. The true definition of a LLM is a type of natural language system, nothing more. For this context let us say AI instead of LLM.
AI is a portion of what you are talking about but not all of it.
Repositories and CI/CD are another subject in addition to the previous two. In the past there were source change control systems, deployment (promotion) systems and other things. Systems like Git have combined multiple things into one system that developers are familiar with.
There is often no need for creativity in many things developers do. We could say that AI gives developers the freedom to be creative when needed or when possible without the need to spend time on the mundane.
Yeah, I get that. It used to feel like art — experimenting, building things from scratch, and learning along the way. Now it’s starting to feel more formulaic, like we’re just optimizing processes instead of creating something meaningful.