I thought it would be fun to make some predictions about the future of development in 2025, and look back in a few years’ time, especially with all of the AI things going on. This is probably naive and wrong!
It’s okay, for now
At the companies I’ve interned for, they usually don’t care much about AI at all, or disallow the use of it. Companies implement rules that disallow AI use for proprietary reasons. I’ve heard of other companies that don’t have as much of an issue with AI, and it seems like in academia, just about everyone uses it to some capacity.
Overall, it seems like the generative AI hype has plateaued in recent years. It is useful, but it isn’t as useful as we thought it would be, and it is not good enough to replace human thought and decision-making yet. So, my argument is that it’s OK for now. Jobs are still available, and we haven’t been completely replaced in terms of code yet.
So, with that in mind, here are some predictions still centered around AI and the software engineering industry.
Focused around code, but less coding
My friend Sumner sent a good quote out recently to a group of computer science students:
“Reading code and understanding it quickly is going to be the main skill of the post-LLM developer’s toolkit. Currently the best way to learn how to read and understand code quickly is by writing lots of code but I think that it’s possible that there will be people in the future who suck at writing code but are good at reading it.” - Sumner Evans
I think this hits the nail on the head. The future of software development is around code, as always, but coding is not the only thing that is asked of us anymore.
Programmers have always had to read and understand code quickly, but this is now even more vital, especially when AI generates bad code that’s hodgepodged together.
Furthermore, code isn’t going away anytime soon. Even if, in the worst case scenario, developers are reduced to just maintaining bugs in large systems made by AI, we must still make an effort to learn how to code and understand it.
As an aside, I do think this is a sobering and sad reality. Part of the fun of my job for me is the code itself. But that mindset seems to be going away now.
More design
Though it has been attempted, design seems to be something AI hasn’t been able to crack yet. Whether it’s literal designs of how websites and apps look, or design of systems and architecture, this requires people.
..and this requires people, because people have abstract problems that require people to solve. Generative AI is just a probability machine; perhaps it can guess correctly at a design, but it usually won’t create something maintainable far through the future that fits the needs of a client.
Software developers will have to learn how to rapidly iterate and tweak their designs to fit a need. The code is becoming easier and easier; problem-solving and design should be the same as ever.
Design is innately human and innately creative. To engineer is to design.
People
I would say “more people” like I said “more design”, except it seems like the “people” side of software development is left to product teams, implementation specialists, tech leads, and others today. But, in a way, I think software developers will become product teams in the future.
Us engineers will be challenged to interact with clients more, truly listen to them, and develop a plan for how to implement code that solves a problem. Even if AI can write the majority of our code, we need to be the glue between the code, the people, and the problem.
I would go as far as to say that academia should teach this just like they teach how to code in software engineering. Academia should heavily emphasize communication skills and working with clients and people directly.
The stereotype or the mindset of computer scientists being shy, or not sociable nerds will (or should) probably go away. There will always be a demand for engineers, and perhaps a growing one, for those engineers who can lead people and interact with them on top of their normal responsibilities.
People are the one commodity that will never go away.
Business value
At the end of the day, companies will be looking for people who add value. Whether they add value in terms of their personality, their ability to write code, talk to people, and so forth; people make the world go ‘round.
As long as it adds value, it’s beneficial to a company. Value makes money.
Computer scientists have traditionally been asked to add value in terms of the code they write; but now they are being asked to do so much more. We will be tested, but I think it is for the best.
Generative AI may be frightening for our jobs and the coding side, but the design and people sides of computer science aren’t going away any time soon. We must add additional value aside from just what an LLM can produce; that’s the future!