AGI won't create new jobs and here is why

If we define AGI as something that performs as well as humans on all current economically valuable tasks, then it could theoretically be true that new tasks will be created that the AGI is not good at, which humans could then make their new niche. In the following argument, I'd like to show that it is possible and likely for AGI to replace all jobs and future jobs (at least for the jobs where success is measured in productivity/quality