So since AI is going to do most of our work and soon all of our work as software and web developers and engineers.
What is the point of keep learning the courses on free code camp?
Any good orientation so we can build our skills better and be efficient instead of replaceable?
I love this sort of question, because initially it seems like that’s what will happen. But at the same time we only have to look at other industries to see how strong automation actually ends up used. My favorite is aviation.
If you board a commercial flight and fly somewhere distant. 90% of that flight is flow on “AI”, of course it isn’t some LLM, its something similar but the same idea applies, its automation taking “jobs away” from the pilots. However, automation is not used or even available in critical periods such as take off, taxing, or landing. It also isn’t responsible for any number of “non-flight” related activities, such as flight planning, briefing and communicating. Its physically possible for all of this to be handled by automation, but it isn’t day-to-day.
Now you might ask why? Why not save money by having automation fly every flight? The answer is simply it can’t reliably do that at the scale and capabilities of a well trained crew with automation. Pilots use autopilot to reduce workload, and using these tools where appropriate. The bar for commercial aviation is extremely high, because lives are at stake, the entire sector has a high bar on safety, you wont fly a plane with less than two qualified pilots for redundancy. If AI flew the entire flight gate to gate it could probably perform well in 99.9% of cases, that’s literally thousands of deaths every day, compared to literally 0 with humans in the loop.
Pilots are often said to make their entire paycheck for the entire year in a few critical moments when things go wrong. You can’t rely on autopilot to solve these problems, (heck there could be a problem with the auto pilot) you have to rely on training, experience and raw flexibility.
So this brings us back to AI and development. Development isn’t usually as life and death as aviation, but it also is more complex than flying from point A to point B.
AI is very flexible, doesn’t get tired and is capable of solving all sorts of problems. But its flawed in its concept of what the problem is due to lack of context. It over-engineers because that’s all it can do. It assumes because its design to make its best guess, not do the right thing. It isn’t actually that smart, it only appears smart. That doesn’t make it any less useful as a generic tool that can be applied to any sort of pattern related problem, but it still has limitations.
No amount of technological progress changes the fact the model you are using doesn’t even know what day it is without calling some external system, it just knows when it was trained. It can’t “learn” anything new without relying on local database knowledge (RAG) that could be shotty at best. It can only spit out its best guess of what to do next based on an input. Everything else it does or leverages is actually good-old-fashion code.
So where do you, the human developer come in?
You have to know how to use AI, while also being able to understand its limitations, and where it “has gone wrong”. If you are not experienced enough to spot when an AI makes a mistake it will appear like an unstoppable knowledge force, but that’s more to do with lack of experience than it actually being perfectly knowledgeable.
You as the developer have to get to the level to spot the problems, understand the issues and take over from AI when it goes wrong. How you get there is harder today, as there’s always the “easy way out”. But this has always been somewhat available due to being able to search online for help. Now its just more accessible with AI tools, but you must avoid them during learning as much as possible as to learn as much as possible.
In closing, AI is here to stay, but to actually replace people completely means leaving it to do all the critical work its completely bad at and then seeing it explode with no fallback. If you have a professional who knows how to use it, they can leverage it for all sorts of things, while still being good at the “human things”.
Good luck, keep learning, keep growing and keep asking questions!