The interest to have AI tools was always there, but the technology wasn't advanced enough.
15 years ago when I worked on my diploma project, having handwriting recognition on a mobile device where you can draw letters and have a CNN (Convolutional Neural Network) was considered a pretty novel usage of AI. Nowadays it is everywhere.
AI tools like Claude and Codex help developers to be much more productive (GitHub Copilot claims 2x). It's also more fun, because the road from idea to a visual product implemented even for a proof of concept is much shorter.
There are two schools of thought on AI's impact. The first warns about the negative consequences for society. The second one sees AI as an amplifier - a tool that helps us be more productive and achieve more.
I lean towards the amplifier view. Let me explain why.
Augmented not artificial
Recently I come accros to a study that stated 92% of daily AI users report felling more productive. Yet only 56% of CEOs say they've goten nothing back from their AI investments.
This gap is where most AI adoption goes wrong. And it usually goes wrong before anybody is touching the keyboard.
The question we need to answer is: what problem are we actually trying to solve?
Not "how do we use AI" or "which tool should we adapt", but what is broken, slow or expensive right now and would AI change that?
I learned to think about this through the 3T framework. When you want to aplly AI to a problem approach it with the following points:
- Time.
- Trust.
- Transfer.
This can be translated in efficiency, quality and scalability.
- Efficiency is speed. Does it go faster using GenAI? Did you spend as long checking the output as you would have spend doing the actual work?
- Quality is trust. Does the output hold up? Are incidents created or the number of bugs increased significantly?
- Scalability is transfer. Can others use it or it is locked inside one person's head?
The goal isn't to optimize one corner. It's to move all three without breaking any of them.
Artificial Inteligence replaces, one the same time Augmented Intelligence extends. When a team gets this right, AI can make individuals more creative and more capable.
Researching I discovered that the move to AI for teams should be done through three stages. First they invent, using AI to do things that weren't possible before. Then they innovate, improving what already exists. Then they grow with AI embedded in how they think, not just what they build.
But, what about if my team still debates about the adoption? It will make us more efficient and then be replaced?
We can apply the Jevons Paradox to this rationale. When software becomes cheaper and faster to build, the demand expands to fill the gap that efficiency creates.
It doesn't look like fewer engineers. It looks like more engineers, working at a higher level, on a different problems that weren't worth solving before. Routine work will slowly disappear and new ype of work will open up.
This is why I'm thinking "AI will replace developers" misses the point. The question is in fact, what kind of developer do you need to be now?
Recently I found this quote that goes like this: "If you're chopping wood with an axe and someone offers you a chainsaw, don't say you're too busy to take it."
I think AI require us to stay curious, think critically and be willing to pick up the new tool.