These kinds of comparisons rarely lead to good discussions.
Let's instead be focused and talk about real stuff.
Consider https://learnpythonthehardway.org/ for example. It has influenced a generation of Python developers. Not just the main website, but the tons of Python code and Python-related content it inspired.
Why would anyone write these kinds of textbooks/websites/guides if AI can replace them? AI companies are effectively broadcasting you don't need the hard way anymore, you can just vibe.
Arguibly though, without the existance of Learn Python the Hard Way and similar content, AI would be worse at writing Python stuff. That's what I mean by "main source of food", good content that influences a lot of people. Net-positive effects hard to predict or even identify except for the more popular cases (such as LPTHW).
If my prediction is right, no one will notice that good content has stopped being produced. It will appear as if content is being created in generally the same way as before, but in reality, these long tail initiatives like LPTHW will have ceased before anyone can do anything about it.
Again, I don't see a way out of this scenario. Not for AI companies, not for content writers. This is going to happen. The world in which I'm wrong is the best one.
In a similar vein, I remember people advocating for replacing new untrained hires with AI. After all, a competent senior engineer is needed to validate the contributions of the new hires anyway and they can do the same checking the AI code.
But then, how would you even train and replace those competent seniior engineers that do the filtering when they retire? The whole system was predicated on having a chain of new hires that gain experience in the process.
From what I could perceive, companies believe coding AIs will eventually learn to both code and teach better than seniors.
This is based on two assumptions:
- AI will get better. Developers using the system will transfer their knowledge to it.
- Seniors in a couple of years will be different. They should be those who can engage with the AI feedback loop.
Here's why I think it won't work:
- Senior developers learn more than they can produce. There is knowledge they never transfer to what they work on. Internalized knowledge that never materializes directly into code. _But it materializes indirectly_.
- Senior developer knowledge come from "schools", not just reading. These schools are not real physical locations. They're traditions, or ideas, that form a very long tail. These ideas, again, are not directly transferrable to code or prose.
- Juniors get embarrassed. You say "stop making this nonsense", and they'll stop and reflect, because they respect seniors. They might disagree, but a pea was then placed under their matress, and they'll think about "this nonsense" you told them to stop doing and why. That is how they get better. So far, AI has not demonstrated being able to do that.
The production of quality content is an aspect of one of those "schools of thought". You are supposed to bear the responsibility of passing the knowledge. Keeping lean codebases easy to understand is also a hallmark of many schools of thought. Working from fundamentals is another one of those ideas, etc.
Let's instead be focused and talk about real stuff.
Consider https://learnpythonthehardway.org/ for example. It has influenced a generation of Python developers. Not just the main website, but the tons of Python code and Python-related content it inspired.
Why would anyone write these kinds of textbooks/websites/guides if AI can replace them? AI companies are effectively broadcasting you don't need the hard way anymore, you can just vibe.
Arguibly though, without the existance of Learn Python the Hard Way and similar content, AI would be worse at writing Python stuff. That's what I mean by "main source of food", good content that influences a lot of people. Net-positive effects hard to predict or even identify except for the more popular cases (such as LPTHW).
If my prediction is right, no one will notice that good content has stopped being produced. It will appear as if content is being created in generally the same way as before, but in reality, these long tail initiatives like LPTHW will have ceased before anyone can do anything about it.
Again, I don't see a way out of this scenario. Not for AI companies, not for content writers. This is going to happen. The world in which I'm wrong is the best one.