You know what a knowledge silo is and if you don’t, look it up. There are hundreds of articles that can explain it better than I can.
One quick search later…
Now that we are on the same page, we both agree that knowledge silos are bad. We’ve been trying to break them down since the inception of software engineering. Why, we have so many processes in place for this that it’s practically a science at this point.
Yet here we are, lovingly creating new ones in the face of LLMs and we are perfectly fine with that!
What happens when they are not there? What happens when companies get dependent and the AI providers decide to exponentially increase their prices? Our planet’s resources are not infinite and as it is LLMs are consuming the at a rapid pace. Can we really do meaningful progress when we are struggling as is? We can’t scale forever.
Training AI on company code and processes is slow and expensive. Not to mention when the next big model comes (that took resources worth billions of dollars to train) comes along and you have to now re-train the model.
I wonder if we spent the same amount of time and resources helping, teaching and developing Humanity, where we would be? It’s not efficient? Dear reader, you are a human.
Now, what I fear is how lazy we've become. The next generation of devs are not trained on Best Practices, Bad Practices, Performance considerations. They are not interested in the intricacies of systems and the origin of it all. There is an Over Reliance On AI, that is replacing Critical Thinking in favor of Speed. The aftermath will be spectacular.
Vibe Coders
Let AI develop AI!