Despite all their problems (hallucinations, confident mistakes, etc, etc) current LLM AIs could be all the AI needed to make capable Jetsons-style robot maids. No waiting around for AGI, as they won’t need it.
Why?
It turns out that with enough exhaustive training data, LLM AIs are very good at learning narrow tasks. Robot Maids don’t need to be able to handle every situation that might happen. They just need to be good at the Top 20 Household Tasks (1. Clean the bathroom, 2. Hoover, 3. Fill & Empty the dishwasher, etc, etc)
There are numerous teams around the world working on this approach, and Weave Robotics has taken the same approach. Isaac 1 is a US$7,999 one-time purchase US$449 per month subscription. For that, it promises to deal with the laundry & tidy rooms back to their original states.
This soft robot wants to fold laundry without pretending to be human
Saved you a click: No.
The realities of consumer robotics is the limitation of consumer motors. No humanoid Robot, no matter how good the training data, is ever going to come close to human speed, dexterity, and importantly our versatility.
This has been obvious with ever single humanoid robot soccer game, mma, and dance competition from ever single highly advanced robotics research group. And those aren’t “cheap” consumer bots. Those are bespoke bots with a singular job.
We should stop trying to hype up Jetson ass maid robots. They only exist in Asimov’s dreams.
Also legitimately, hiring a human maid once a week for an hour for like $50-$100 is the correct answer if you’re set on not folding your lay dry and you haven’t accepted the permanent pile.
