We are 5 years in... it's fine to be sceptical. The model advancements are in the single digits now. It's not on us that they promised the world 3 years ago. It's fine and will be just fine for the next few years. A real breakthrough is at least another 5 years away and if it comes everything you do now will be obsolete. Nobody will need or care about the dude that Sloperatored Claude Code on release and that's the reality everyone who goes full AI evangelist needs to understand. You are just a stopgap. The knowledge you are accumulating now is just worthless transitional knowledge. There is no need for FOMO and there is nothing hard operating LLMs for coding and it will get easier by the day.

5 years ago we had GPT-3, not even instruction-following GPT yet, a mere completion model. ChatGPT release was late 2022 (3 years ago). True agentic systems with reliable tool calling in a loop, that came maybe a year ago, agentic coding harnesses less than a year ago.

Model improvements may have flattened, the quality improvements due to engineering work around those models certainly have not.

If we always wait for technology to calcify and settle before we interact with it, then that would be rather boring for some of us. Acquiring knowledge is not really that much of a heavy burden that it's an issue if it's outdated a year in . But that's maybe just a mindset thing.

I haven't been listening to any promises, I'm simply trying out the models as they get released. I agree with the article wholeheartedly - you can't pretend these tools are not worth learning anymore. It's irresponsible if you're a professional.

Next breakthrough will happen in 2030 or it might happen next Tuesday; it might have already happened, it's just that the lab which did it is too scared to release it. It doesn't matter: until it happens, you should work with what you've got.

I would have wholeheartedly agreed with this comment one year ago. Now, not so much.

Where we're at is a lot better than we expected to be three years ago TBH.