When we replace Roombas with Moravec-class androids that perform common domestic chores, then I think we'll really be tackling the paradox. Right now I am guessing we're at the approximate equivalent of perhaps say, just before the early Cambrian Period, when vision systems started to show up in the fossil record.
So I'm hoping for an exponential development explosion that delivers androids that do laundry (sort, wash, fold, put up, identify for recycling), clean and prep food ingredients, dust/vacuum/mop/cleaning in general, return items to their places, make beds, take out and return rubbish and recycling bags and bins, etc. Even with a developed world income well above the 90th percentile, I cannot cost justify spending ~$0.5-1.0K USD for small slices of each of these capabilities with specialized equipment to automate them, which seems to be about the threshhold they're introduced at.