Not fresh news but that’s noticeable that when AI writes code and proposes architecture, you as an “AI orchestration operator” need to sort out what is good and what is not, and to do that you need to have:
* experience in the industry to understand what is working in real life with humans across the product * domain experience to know what problem you are trying to solve and for whom * pattern knowledge from good old books
And the last thing is becoming more and more important each day. Since you can relatively easily understand what is working well and who your target audience is, but the language that really speeds up things and helps you communicate- is patterns.
So instead of daily saying to AI “maximize reusability and make sure that we can easily make changes in the future for feature X” - you say “use SOLID and DDD for module X” and this level of abstraction will allow you both (AI and yourself) to talk about very particular, concrete and well known things instead of speaking as a plain human - use that as advantage not as handicap
PS cat photo for attention 😁