Absent human-level AI, there will remain tasks that only humans can do. Hence we will have jobs.
But truly, if we do get human-level AI, it seems to me we will have bigger problems than unemployment. I hate to get all Nick Bostrom, but it really does seem like the objective functions of the machines would be all that matters in this scenerio. As it is their objective functions, not laws, that would decide our fate. Either human-level AI is far off, in which case technological unemployment is not a worry, or it is not in which case technological unemployment is the least of our worries.