AI has been a household word for what it seems like eternity but for sure the last 3 years. The biggest milestones - depending on when your interested peaked in the topic of generative models, LLM, ML or just AI in general IMO are the following:
- GPT-1 released mid 2018 - early adopters were amazed, researchers baffled and OG - the best dota team in the world: confused
- GPT-2 released early 2019 - many have started building zero-shot prompts and utilizing many benefits from this model and this was the next step in gaining popularity but not productivity. Not yet! At this point many have started talking about AI and the "wonders" and "doomsdays" of this technology
- ChatGPT released late 2022 - a full-blown momentum shift for the whole world, the fastest growing userbase in any piece of software and definitely a household name within less then a year.
In the meantime it's worth noting that other companies other then OpenAI have joined the race with their models and innovations, but still there wasn't a clear winner. Each model would trump the previous one, each new innovation would be quickly copied and the only clear winner was a company that didn't even build their own models but instead digital shovels. Regardless, at some point many have started questioning the value of these models in generating images, videos and the "easiest" task: text.
But not just any text, the one text that's not readable to most of the planet (0.3721% of the population) - yes, software code. And this type of generation was easy to train as most of the knowledge lives online, and it's mostly free and accessible. Examples? Yes, Github and any other platform hosting code with public (and maybe private?) repositories was a contributor in training these models.
So - can we enable software engineers with this tool? Can we help newcomers start? Maybe even non-technical people to get into software engineering? Or just replace all of them with agents that write code on their own. Many disagree on answers. CEOs say models will replace people. Bots/models/agents/AI (call them however you like) will replace faulty humans that need vacations, sick leaves, mistakes to learn, time to eat, time to sleep, time to breathe air.
The Reality
The truth, as always, lies somewhere in the middle. AI is not replacing engineers - at least not yet. But it's definitely changing how we work. The engineers that embrace these tools and adapt will thrive. Those that refuse will struggle to keep up.
I've been using AI assistants in my daily work for the past couple of years. Here's what I've observed:
-
Boilerplate is dead - Need a basic CRUD endpoint? A standard React component? A database schema? AI generates these in seconds. What used to take hours now takes minutes.
-
Learning is faster - Stuck on a problem? Instead of scrolling through StackOverflow or digging through documentation, I can ask an AI to explain concepts in my specific context. It's like having a patient senior engineer available 24/7.
-
But the hard stuff is still hard - AI struggles with complex architectural decisions, nuanced business logic, and problems that require deep domain expertise. It can write the code, but it often doesn't understand the why.
-
Bugs are still bugs - AI generates code that looks correct but has subtle bugs, edge cases, or security vulnerabilities. Someone still needs to review, test, and validate everything.
The Future
I don't think AI will replace engineers. I think engineers who use AI will replace engineers who don't.
The role is evolving from "code writer" to "code architect and reviewer." The value we provide is no longer in typing out syntax - it's in understanding problems, designing solutions, and ensuring quality.
For newcomers, AI is a double-edged sword. It makes it easier to get started but harder to deeply understand what's happening under the hood. The temptation to copy-paste AI-generated code without understanding it is strong. And that path leads to being a junior developer forever.
My advice? Use AI as a tool, not a crutch. Let it handle the repetitive stuff so you can focus on the creative, challenging work that actually moves the needle.
The future is not AI vs Engineer. It's AI with Engineer. In most cases. In some it will be Engineer with AI. But it's unlikely that it might be one or the other.
I'm not sure that a complete swap will happen in the near future. Maybe not even in our lifetime. Humans are still amazing, creative, and so much more and these traits spark thought and innovation - which is the reason humans had so much progress even before the AI era.