Software Development and AI: A Musing

Fear of Software Evolution

There's a murmuring of AI "taking jobs" from software developers, graphic designers, musicians, and the like. Whether it's GPT-3, GPT-4, or the next iteration of machine learning, one thing is true: it's evolving.

This is not bad news. There's much fear associated with these breakthroughts; fear from developers and designers of all levels. Many are worried that the new technology will replace them or make their creative problem-solving skills obsolete. New developers may be concerned that their chances of breaking into tech are shrinking slimmer by the day.

I hope this musing helps allay these concerns.

Embracing Change

Things are changing. Fast. Within the last decade, artificial intelligence (AI) and machine learning (ML) have taken tech by storm. It started small and relatively unpromising. However thanks to relentless human creativity and drive, it grows exponentially more sophisticated and reliable.

This should be no surprise! People have created incredibly complex and sophisticated technical advancements for centuries. This is not the first technical revolution, and we can learn from the past.

Consider: Algorithms

Consider the state of the current technical interview process. It's not uncommon to have a developer jump through hoops programming algorithms, demonstrating their in-depth understanding of data structures, or narrating their live coding solutions to some arbitrary logic problem... all to gain the trust and acceptance of their interviewers. Why? Most algorithms have been solved; their speed and efficiency is understood. Why do we constantly need to rewrite them or reiterate our understanding of them?

So, what if AI coding could handle all the algorithmic solutions? What if it could provide the correct data structure for the problem at hand? This would mean that software developers would no longer need to spend hours working on such things. Instead, human developers could spend their time piecing together the manufactured parts of a larger application, fine-tuning performance, user experience and developer experience.

Rather than spending hours finding the most performance algorithm to sort a set of data, we could prompt AI: "write a function to sort a data set where items are provided in this structure: [structure]. The sorting function should be as fast as possible." ...then we can shift gears to work on the more complex parts of a project — the parts that require human intelligence.