A Realistic Take on the Future of Programming With AI

Photo by vackground.com on Unsplash

AI is still rather a hot topic for software engineers. In my company, there is still chatter about whether we will have jobs in a few years as the use of AI becomes more widespread in software development.

Here The Secret Developer presents their take on what the near future might hold for software engineers who choose to stay in the field.

The Current Picture

Right now software developers should not be worried about their jobs. As I’ve previously written, AI is currently coding standards and we need real people to sort through the copy-paste generated code and actually give us maintainable software.

That’s going to keep us all in jobs for at least the next 18 months to two years. Then, the following may well start to happen.

The Predictions

You’ll need to go fast, and have No Place to Hide

AI will take on the brunt of basic coding tasks, and employers will crack the whip on delivery times. 

Well-defined and well-written requirements can be burnt off with little to no developer intervention. This will leave just those tricky bugs and difficult-to-implement features, which might mean I’ll actually have some fun work to do.

Those who padded their timelines excessively will be eviscerated. AI will also give accurate estimates for the delivery of features, setting productivity and quality standards.

This means that some of my colleagues who deliver next to nothing will be out of jobs. Nothing would be lost here.

Quality Police

Although I think standards and quality is desperately needed, automating this will lead to some hard falls for lax developers. 

This won’t just be SonarQube and clicking “won’t fix”, it’s about having an AI look completely through your work and question why you aren’t doing everything better. It’s not going to listen to you explaining you have a tight deadline either.

Linting and vetting of code are likely to become part of your pipeline, with the only human intervention being a senior giving a sanity check.

Your local linter as of 2024 will be retired completely as it simply cannot cover all of the cases as well as the AI linter of the future. Those who cannot keep up to the exacting standards are likely to be phased out too, like a university student handing in a paper with spelling mistkaes after the advent of spell checkers.

This might mean we get some maintainable code, which would actually be nice when implementing features (I think!).

Depth is valuable

With AI covering the breadth of technical know-how, deep expertise in critical areas will become more valuable. Mid and junior roles will become more difficult to find as only senior quality is sought by companies. 

The T-shaped developer is likely to be edged out. AI has a broad but shallow knowledge base that makes this type of developer next to irrelevant in the future.

Agile changes

AI tooling will spit out estimations and plans. We will see AI producing fancy burndown charts, and even explain what it means to the business (in a way even the business can understand).

This one will probably benefit everyone, Agile coaches will be able to focus on people and communication. There won’t the current pattern of nobody paying attention in stand up meetings, thankfully.

Conclusion

“The future is not set. There is no fate but what we make for ourselves.”

I guess that means we better start delivering the features we promised for next week, then.