GenAI has introduced the greatest level of abstraction we have ever seen in software development.
If there’s anything that has become clear to me while using GenAI, it’s this:
AI speed is abundant. Craft is scarce.
Those inexperienced in software development while embracing GenAI may never understand how software is best constructed.
It’s the same kind of challenge that occurred by using an IDE but not understanding how compilers and other tools worked.
Even worse is having no knowledge of how computers and their peripherals work.
As a result, when using GenAI, features will be continuously rolled out and attached to freak, ghoulish applications and services that become part of ignominious systems.
Let me be clear about what I’m saying here, because I know this will be uncomfortable for many of you.
What I’m Not Saying
I am not equating lack of knowledge about compilers and hardware with a lack of knowledge in constructing software. What I am saying:
- The higher the abstraction, the further away you are from the real thing
- The potential for GenAI to dumb-down the industry to the point of complete lack of caring at all is at the greatest level we’ve ever experienced
Likewise, I am not saying to throw out GenAI and other generative tools. I’ve been building such tools and benefiting from them for four decades. What I am saying:
- With the right knowledge and care, a single engineer can do amazing things
- Yet, that doesn’t happen without struggling with your senior-junior-senior pair programming partner in a box; that is, your GenAI coding assistant
- You’d better know your stuff
Note the words “lack of knowledge,” “engineer,” “struggle,” and “know your stuff.” Those words were not chosen by GenAI. They were chosen by me.
And by “know your stuff,” I definitely don’t mean:
- technical mechanisms, such as databases
- cloud APIs
- frameworks
- “Agile” certifications
- programming language certifications
It’s not wrong to have those skills. Any of those could be useful. Yet, they are learnable skill sets and those are all the areas in which GenAI already excels.
My Uncomfortable Prediction
Some software development skills will be made irrelevant. I’m calling out this big one right here, and I think this is more fact than prediction:
- Frontend developers are the first to go
I don’t wish that on anyone. It’s a simple fact that I have experienced while using GenAI. This is a conclusion that I have drawn:
- I will never again hire for frontend development
My opinion is that this will become clear to many hiring managers. If you face that situation, here’s my second prediction:
- You’d better grow
The question is: How?
The Path Forward
We’re at an unprecedented crossroads. GenAI is not going away. You have two choices:
- Resist GenAI
- Embrace GenAI
I have seen numerous technologies come and go. Along with the obsolensence, I’ve seen relevant skills become irrelevant or commodity at best. Too many people became too comfortable with what they knew. In the fast-paced software industry, that generally hasn’t gone well.
Don’t be that person.
The question isn’t whether GenAI will change software development. It already has. The question is whether you’ll change with it.
Learn the skills necessary to remain important to the business. Those skills are not available in GenAI. That’s because they are not easy to learn from a web of online documentation.
You’ll need to decide whether learning Domain-Driven Design and advanced architecture patterns will make a difference in your skill set. I predict they will.
