One thing junior devs at my company are taught early and often is that it's OK to look at a potential solution that an AI could come up with, just like it's OK to use stack overflow, but to never ever trust that solution without fully understanding it and usually simplifying it quite a bit.