Decided to use Deep Research to gauge the progress of domestic robotics.
https://chatgpt.com/share/681e9890-390c ... 450bca91a1
My initial prompt:
Research the latest in AI and robotics. Recall a post I made earlier (in ChatGPT) about neurosymbolic AI and backpropagation and tree search and agent swarms.
My question to you is, just how close are domestic humanoid robots? Be brutally honest (and by that, if brutally honest means they're imminent, that's also good; if they're a long ways away, that's also good). The cost isn't necessarily important, but it's welcome to see how these things progress. It seems China and the USA are leading the pack, but I have a hunch China's actually about to pull ahead.
With AI, recall the concept of universal task automation
> universal task automation. It would be simpler on everyone if we based definitions of AGI on actual task capabilities rather than more abstract "as intelligent as an adult human" or "capable of recursive self-improvement" (because even a narrow AI could theoretically be capable of the latter if it was purposed for it) This is one of my "working backwards" definitions— an AGI should be capable of universal task automation, meaning any individual task (unit of work) is automatable. The cost is not important. Even the proof of physical ability is not important, long as any conceivable task can be done by the AI if given the capability. All jobs and hobbies are composed of smaller units of labor, tasks, so universal task automation essentially means all jobs are automatable, but not necessarily that all jobs will be automated.
Solving machine vision and commonsense reasoning are obviously very important towards solving robotics.
So to provide a few core research topics:
- How close until we have a domestic utility humanoid robot for sale? A utility robot in my definition is the classic concept of a home helper servant robot able to do any task requested (If we already do, in limited manner, great!). You can take time to bring up the phenomenon/grift of social robots in the 2010s, seeing as that gave us two humanoids (the toylike NAO and the more reasonable Pepper), while also explaining the obvious about why the 2010s social robot fad never went anywhere (i.e. they were based on traditional chatbot technology, typically built off Markov Chains and very occasionally LSTM models, so the "social" aspect was severely limited compared to modern LLMs)
- What needs to be solved to put a humanoid robot in an average middle class home? (Hard mode: no stopgaps like wheels or tracks, the consumer likely prefers legged bipedal humanoids) What does AI need to solve? What is limited us in terms of hardware, like battery life or mechanical costs? Assuming we're an eccentric billionaire, millionaire, hundred-thousandaire, and then eventually under-$100k a year (payment plans and insurance ala buying cars could still make it affordable), what's the likelihood of getting one if I wanted one right now, vs 5 years from now, vs 10 years from now, vs 15 years from now?
- As a bonus, what's a possible expected timeframe for realistic androids/gynoids? Taking into consideration hyperrealistic art and sculpture, the raw appearance has already long since been solved, but convincing mimicry of human motion and expressions seems to be a long way off. We can make hyperrealistic static robots, but the moment we set them in motion, the Uncanny Valley sensation triggers harshly. Is there any progress towards overcoming this?
- Another bonus, what would be some side effects of domestic utility robots being commonplace? Consider my hypothesis on "Technism" and the "helot" concept, which in itself was forged by keeping a journal and writing my daily life with the daydream of it being augmented by advanced technology, and coming to the conclusion that "renting out" a domestic robot when unused to the city or state or community could create a sort of modern helot-like service of robots doing community services and self-repairing cities (and then realizing that the city/state/community could simply do this outright). This is something that is not immediately obvious when thinking of utility/service robots, but becomes obvious and inevitable once conceived. Anything beyond that?
1. Include all known robotics projects
1.a: All humanoids are fine, but the hard mode goes that the average consumer would prefer bipedal humanoids if they're available (also most human spaces are unlevel to some extent, which might also provide incentive against wheeled humanoids being more common as it'd just be safe to consider that an average middle class home or apartment would require a robot to navigate steps or uneven surfaces that could cause a wheeled robot to tip over, but it's still a good idea to consider wheeled robots too)
2. Yes
3. Both, entertainment and companionship (and let's be completely honest, sexual purposes) are going to be one of the biggest markets for robots, especially lifelike ones, but utility and service help is obviously necessary
4. Yes
