GPT technology, exemplified by the Github Copilot and its likes, changes software engineering to the ground. There is no doubt that the technology places a new tool in our engineering shed. It allows us to create software with a completely different set-up than what we are used to.
Now, what it really changes is only a few things, but these are very big ones.
- Programmers —> designers and architects. GPT can write source code like no other tool on the market. And it only gets better at this. A quick glimpse at the Github Next website gives us a good understanding that this team has only got started. This changes everything we know about engineering software. Bad programmers will disappear over time. Good software designers, architects and software engineers will take their place. They will be fewer in number, but better in quality.
- Software development —> software engineering. Designers will no longer get stuck in solving a small bit of a puzzle. GPT will do it for them. Instead of thinking how to write a test case, the designers will think how to test the software in the best possible way. They will focus on the engineering part of the software engineering. Something that I’m teaching my students from day one.
- Consultancy —> knowledge hubs. Since programming will become easier and more approachable, we will need people who know how to solve a problem, not how to write a program. This big chunk of business of the consultancy companies will disappear. The consultancy companies will specialize in their domains and in problem-solving.
There will also be other things that will happen. Requirements will not be the same as they are. Testing will be different, architecting will be smarter and management more optimal. Knowledge will be more valued and critical thinking will be needed even more.
Well, this is my end of the academic year blog post. More to come after the summer. Stay safe!
