Tim Cooke wrote:On AI, I have a relationship with it based on caution and scepticism. It presents solutions that look good superficially but rarely are good and often require considerable rework, sometimes more rework that I'd have expended just doing it myself from the start.
I'm with Tim on this. We even had a face to face chat when we met few months back about AI and how it shapes the industry. While you could see some value it brings, but it also brings a lot of problems you spelt out already, and future problems we are just about to see later.
One of those problems is, engineers, or rather those who are meant to become engineers (think students or junior developers), becoming so reliant on AI, that is the only thing they know and blindly follow. Instead of thinking and researching (for hours) and coming up with some sort of approach in mind, and only then asking for certain help to get there, they instead asking
"how can I do it?". And now, you may see the way "how" to do it, but the next time you come to a problem, what's the first instinct? Let's ask AI, because it
told last time.
If I'd compare it to a problem, i.e. you need to cross the city. You could jump on an underground train and get from East to West in under an hour, or you could jump on a bicycle and do the same, surely it will take much much longer. However, you can imagine how much you'd learn about the city while you cycling, its streets, road conditions, traffic it has, density of the people, variety of shops around, ALL SORTS OF THINGS you don't need to know IF ALL you had to do is to cross the city, BUT, do you see how much knowledge you would have built that way? So the next time someone asks to solve a problem how to build a road to cross the city from East to West - you'd know where to start, you'd have something in your head, called knowledge/experience.
And to obtain such experience ^ there are no shortcuts as to just build it in think habitual, old school way. At least I don't know currently. Maybe in N years the abstraction layer will get lifted, and writing code and solving problems the way we solve today won't matter, I don't know, but what I see and experience myself from the surroundings, AI just dulls the natural engineering senses.
So, try to balance. It is really so fun to think and sleep on a problem until you finally get a crack. You don't need to rush things.