AI and the Future of Software Developers: Confessions of a Code Monkey

We're all done for!

Clickbait worked, didn't it? Great.

Here are my personal insights after nearly a year of using ChatGPT and other LLMs in my coding. Disclaimer: These views do not reflect those of my current employer.

About a year ago, I started exploring ChatGPT. I was a bit late to the party, initially spurred by the somewhat hysterical posts by marketing folks about ChatGPT on LinkedIn. I'm not new to AI, though. Around 2002, I dabbled in building a neural network for fun but was thoroughly disappointed by the capabilities at that time. The effort to build an NN for just recognizing a specific type of image was enormous. Since then, I've occasionally incorporated some AI libraries, or those claiming to be, but nothing truly impressive. AI remained a fringe topic for me until ChatGPT.

I approached the ChatGPT hype with skepticism. Boy, was I wrong. I was blown away by its quality and capabilities. In two weeks, ChatGPT transformed my work habits. Google? Obsolete. StackOverflow? Obsolete. Tutorials? Unnecessary when I now have a Pair Programmer who never complains, is always available, designs my architecture, writes my code, creates classes and methods, debugs, writes unit tests, and even helps me convert Java or TypeScript code to Python, despite not having used Python in over a decade.

Besides programming, I've explored topics like agents, agent swarms, and various LLMs like Llama, Mixtral, etc. Now I have my personal advisors:

- A news reporter giving me daily briefs

- A travel planner for my eccentric vacations

- A chef familiar with my love for Italian, Georgian, and Asian cuisines and my gluten intolerance, and catering to my daughters' dietary preferences (one vegetarian, the other a ramen fanatic)

- A cyberpunk enthusiast with whom I share dystopian visions and design my cyber world

- And of course, my Pair Programmer Buddy, with whom I've developed a lot of code.

Crazy, isn't it?

I've also pondered the broader implications for my profession and society at large. One thing is clear: this shift will affect many professions, especially consultative roles, reducing workforce needs with all its positive and negative consequences. Am I scared? A bit. Curious about the future? Absolutely.

Here are some predictions for software development:

Accessibility of AI: AI will soon be accessible and usable by everyone. Setting up a local GPT/LLM has become trivial. The potential is limitless with free LLMs, datasets, and company-specific data. Minimal programming knowledge and a basic understanding of LLMs and APIs are still required, but this will soon change, democratizing AI use. The implication: nearly every product, digital or physical, will incorporate AI, reducing the need for software developers.

Relevance of Programming Languages: The importance of specific programming languages will diminish, as will the need for traditional programming skills. Programming will evolve from loom cards and machine languages to natural human languages as the default. If everyone can speak or express themselves in some form, traditional programming skills become less critical.

Process of Digital Product Development: This will fundamentally change. The divide between product owners/managers and development teams will disappear, giving rise to the unified digital product developer. This role will likely require more in-depth knowledge of processes, architectures, platforms, and data structures. The pool of talent will come from product developers, product owners, data analysts/scientists, and software developers. Each group brings its strengths and weaknesses. Software developers, known for their process expertise, problem analysis, and focus on essentials, will be invaluable.

The transition to this new paradigm may take about five years, and another five for developers to become obsolete in their current form. I'm excited to witness this change in my career. Let's face it, fellow programmers, our job involves a lot of redundant and detailed work. The future will change this.

Yes, there will still be a need for pure software developers to create the foundations for platforms, swarm assistants, frameworks, libraries, and components. But their numbers will decrease. The rest will evolve into digital product developers, creating better digital products faster and with more value for everyone.

What are your thoughts on the future of software development?