Is it still worth learning how to code if AI will replace developers in the next two to three years as it is improving exponential rate?
Hi! Check this conversation, someone else posted a question that related to this. Let me attach a link here, read through the posts.
I’m not convinced that it is.
I’d hate to tell you this, but AI can never truly replace developers. For all the smarts it claims to have, it still has a really hard time understanding how to solve problems. It can’t figure out how to identify issues in Python code and takes several tries before it even comes close to building something correctly.
And I genuinely doubt is improving at an exponential rate.
You could say that AI will replace any job in any field. Music, art, design, writing etc. It’s like when people thought robots would take everyone’s jobs 50 years ago.
AI is a tool that will no doubt change things, but there will always need to be a human element somewhere along the line. And if it got good enough for that to not be true, then there would be bigger problems than what career you chose.
AI improves at more of a logarithmic rate than an exponential one. Which means it improves very quickly after initial investments, but you get less and less returns over time after the initial jump.
You can see this with current AI use-cases such as AI driving, and text generation.
Where you can get something that works most situations OK, but you can’t seem to iron out the last few kinks no matter how much you keep investing. Anyone who says “it will get better” has to say that as money is being invested expecting it to get better.
More functionality, can be added more linearly, but this isn’t AI this is just old-fashion development work. Prime examples are Bard(Google’s AI) being able to read your emails and tell you about your google workspace. The AI didn’t necessarily “improve”, it just became more useful by being integrated with things. This is all software development.
Modern AI is built on the same theories from 50 years ago. The main difference is the amount of data and computing power available. However, these theories have pros and cons that still are applicable. Unless this changes the same problems we see today will continue.
So this brings us to “what about coders”? If AI will still be hamstrung by limitations, the only solution is to leverage AI as a developer correctly.