Slashdot Mirror
Elon Musk's Open Source OpenAI: We're Working On a Robot For Your Household Chores (zdnet.com)
An anonymous reader writes from a report via ZDNet: OpenAI, the artificial-intelligence non-profit backed by Elon Musk, Amazon Web Services, and others, is working on creating a physical robot that performs household chores. In a blog post Monday, OpenAI leaders said they don't want to manufacture the robot itself, but "enable a physical robot [...] to perform basic housework." The company says it is "inspired" by DeepMind's work in the deep learning and reinforcement learning field of AI, as displayed by its AlphaGo victory over human Go masters. OpenAI says it wants to "train an agent capable enough to solve any game," noting that significant advances in AI will be required in order for that to happen. In May, the company released a public beta of a new Open Source gym for computer programmers working on AI. They also have plans to build an agent that can understand natural language and seek clarification when following instructions to complete a task. OpenAI plans to build new algorithms that can advance this field. Finally, OpenAI wants to measure its progress across games, robotics, and language-based tasks, which is where OpenAI's Gym Beta will come into play.
Can I make one rob a bank as it's win win. I get the cash or I get in to club fed with room / board and a doctor.
When its about AI, open source is not enough in order to let people have an usable product. The open source license only covers the source code, not the training data. These data are much more important, and usually they have very restrictive licenses.
They can't make a robot that's remotely capable of walking normally and opening a door. Not trusting any of that with my dishes. Less even FleshLight.
Would you like to play Global Thermonuclear War?
134340: I am not a number. I am a free planet!
If the cost of the robot is cheaper than the cost of a cleaning lady over the warranty period of the robot, why wouldn't you buy it? If a $5000 cleaning robot was introduced with a 5 year warranty, you'd be crazy not to buy it. That's way cheaper than a cleaning lady and it's something that a middle class person could probably afford and justify. If useful robots genuinely become available to the middle class, it's going to be a huge win for society.
More hipster tech that I can't use. Will it be government subsidized?
This is something you consider when technology is mature. Nothing Musk presents is based on mature technology, and I amso tired of tech that basically just facilitates human laziness. Please stop investing in this junk, billionaires, and please stop allowing our government to change laws to accomodate it. That money could be put to such better use, believe it or not, there are still starving people on earth.
The front page just flies with new bullshit.
GO CIA GO CIA GO CIA
Rockets? Paypal? Electric cars? Residential solar systems?
It's got to be named "Hired Girl, Inc." Yes, it's slightly sexist, but with with all the anti-Islamist sentiment, "Aladdin Auto-Engineering" is definitely out.
To be followed by the latest fat at the gym: dustersize, wherein you carry out the motions of dusting.
That's the basic idea behind kettle bells--you're simulating physical labor in a formalized way. You could spend money on the robot and the gym, or you could just CLEAN YOUR FUCKING HOUSE and GET YOUR FAT ASS MOVING.
Because "sex" is a household chore. (but don't do it in the dishwasher...)
So what we've established here is that you don't know what a monopoly is, and that you also don't know what a trust fund is.
Agreed.
SpaceX? It's effectively a research group that does some real work on the side.
Tesla? Tesla is a an electric sports car company sucking at the teat of taxpayer subsidies and bailouts.
SolarCity? Another company skillfully navigating tax credit, rebates, and subsidies via public funding and extremely strict contracting.
Hyperloop? HA! HA HA HA! ha ha HA!
People need to stop seeing "rich" as equivalent to "genius".
Did anybody notice how AlphaGO's avatar changed in the last game? Was that also a decision made by the AI? ;)
Would you please make a robot in the image and stupidity of Donald Trump: U.S.A. Demogogue ?
Thanks in advance,
K. Trout,
San Francisco, CA
right?
First he's going to get you to pay for an electric car. Then he's going to get you to pay for a household robot. Then he's going to get you to pay for solar panels to power the electric car and the autonomous AI robot.
What do you call a million independently powered autonomous AI household robots driving Tesla cars?
Elon Musk's private army!
Also, I have a newsletter you might be interested in.
---
Ok ok, I obviously didn't have enough random capital letters and exclamation marks to have a newsletter. Still... Robot army! You heard it here first!
Moving the goalposts, I see.
What chores will Rosie do? Cooking? No, way too complex for decades yet: gathering foods from fridge and pantry, opening a variety of containers, exracting indredients of many shapes and consistencies in proper amounts (and avoiding those that are spoiled), prepping each (peeling, chopping, grating, sauteeing), and synchronized cooking before dishing up. All this without burning down the house or spilling and then having to clean up the messes. Not to mention the cleaning up of utensils and pans thereafter.
Cleaning? Not even. Vacuuming, dusting, tidying, navigating a dynamic ever changing floorplan and tabletops, without toppling and destroying all those expensive knick-knacks or running over the cat, and of course, not making an ever greater mess.
Washing the dog? Nursing the baby? Cleaning the windows? Mopping the floors? Beating the rugs? Washing the car? Mowing the yard? Clearing the gutters? Weeding the garden? Trimming the hedge? Edging the sidewalk? Taking the dog for a walk? Taking out the garbage? Yeahhhh...
Which subset of these chores does OpenAI choose for their Rosie Jetson? 'Cause she can't do them all. And how many Rosies can you sell if she costs more than a Segway but only can load the dishwasher and fetch beer from the fridge?
Reality check: an affordable domestic robot that's actually useful is hellaciously ambitious, especially when no robot on earth can do any of these things yet, not at any price.
To be fair, guys working on making working robot hands are not connected with the same groups that are working on bipedal motion and those same groups aren't working on visual recognition, etc. If Musk can unite these different projects he stands a good chance of making a robot for house chores.
We already have robots to wash our laundry and dishes and floors. What we really need are robots that can safely fold clean laundry, stack clean dishes, and load washing machines and dishwashers and take out the trash. Bonus points if they can also kill pests and act as a guard dog.
If someone can make a robot that can clean my apartment good enough for my landlord not to get mad at rent inspections, I will be first in line...
Why does it take so long to make useful robots?
You start with the small tasks first, not robot cars. Roomba had the right idea, it's just not robust enough. Ideally you make something with hands capable of multiple tasks, not specialized robots.
Specialized robots are for businesses who could afford the benefit and costs of such redundancy. Normal people will need robots that can use human tools, not robot vacuums or robot dust mops or robot dish scrubber. Rather... a robot that can use those already existing tools. In the big picture that makes more sense, costs much less and is smarter/safer.
We need a robot that can wash dishes, do laundry and go up and down stairs and only cost a few thousand at most. It should be 100% doable with today's technology and IMO it was doable 10-20 years ago. We have been lazy with robots and waited for battery technology when we didn't really have to.
If we have to engineer wired robots or robots that charge often, we should. The articulation, sensors, and computing power has been around for 10-20 years now. The only thing missing is quality coders. You don't need AI to make robots that pick up dirt and scrub things. You need something even more rare.... a computer/robotics engineer. A real one, not just a person who 'engineeres' network cable and routers. That's not engineering really.
We are relying too much on AI to magically make robotics work. Armies of coders with a understanding of engineering is what we need. We lack coders with much real world physical knowledge and load bearing, hydraulic, and general machinery. Even just embedded coders are hard to find, no less ones who can make a robot with a balanced gait and properly articulating hands.
AI isn't going to do that.. engineers are going to do that, but they haven't and AI won't fix that unless it rapidly become god like, in which case I will hope to be adopted and converted into a digital entity. We are expect a bit too much from the general concept of AI when what we need is more coders and engineers. Robots that can do basic chores, chores that even disabled humans can often do, should have been retail by now.
Motors and steel are not particularly expensive. Beyond that it's mostly a lack of vision and funding. AI is something you upgrade your robots with, not something you need to make them do basic chores. Instead you just make the robot follow commands, like the tool that it should be. Humans oversee it until they are happy with it's in field programming and BOOM, there is your AI.
it's mostly just about making an easy to use UI to tailor the robots scope of work. The idea that AI will come to fruition that allows a robot to just wander around the house doing chores is stupid. That's decades away. We should not be waiting for AI to make robots that can do chores.
It would also help if we stop calling machine learning AI when it has never produced anything like AI. It's more like brute force programming or evolutionary programming. Just remember, 99.9% of the time evolution does not create intelligence. It's a means to efficiency do specific jobs, but that might not be efficient in any way we require in a product. In many cases efficiency is not what we need the most as much as flexibility, which machine learning does not offer.
What would be more useful is user programmability. Let the human be the AI and let the robot be an extension of the humans brain. It's more practical and it would teach EVERYONE programming and problem solving.
It's practical if the owner teaches the robot specific to the environment and tasks. It's not practical if you expect the robot to know how to do everything or just magically learn on it's own like an actual lifeform.
A lot of these tasks are possible if you make them a bit more robot friendly. Like washing windows, it's not too difficult if you design your building to be washed by robots. As a retrofit it might be expensive, but for new high rise buildings it makes sense.
About 80% of vacuuming can already be done by a robot, which saves a lot of human effort. Stairs were a problem but now they are so cheap you can afford to have one for each floor. Rugs don't need to be beaten if you have a good enough vacuum with beater brush. The same tech already exists to mop floors, although you do have to handle the liquids yourself. With a simplified tap interface and cleaning chemical packs that could be overcome.
Washing the car... Robots have been doing that since the 70s at least, when was the first automatic car wash invented? You could probably get 90% of that cleaning done with a robotic arm if you really wanted one at home.
Rubbish disposal seems like a gap in the market. All the components are there, robots that can navigate a home and interact with well defined objects. Price is probably the issue.
As for cooking, we already have pre-prepared meals that only require very simple cooking. Put some stuff in an oven, try a few things, throw it all together. If they could package it such that the oven reads a barcode with cooking instructions, does some basic handling and spits out a completed meal I'd be interested in that. Yes, I really am that bad at cooking.
Musk tends to look at problems this way. Auto-pilot in his cars isn't really a full "robot car", but it does a lot of the work and is improving. When he talks about AI, he is probably thinking of new ways to monetize the AI developed for his cars and future expectations of what it will be capable of.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
I would give Rosie extra oil for dinner if she managed to run over the cat.
is paved with good intentions.
It won't be the first time for Mr. Musk. Lest anybody forget that he is the man responsible for creating Paypal. He did it with the best of intentions.
Now he wants to create AI. I've seen that movie, many times. When he succeeds, and his creation's descendants are travelling back in time to exterminate the human race, we must remember that they were created with the best of intentions.
No, the goal posts are always there: You get praised when you successfully accomplish the goal you say you're going to accomplish.
SpaceX said they're going to do commercially viable space travel and cargo delivery. Still workin' on it.
Tesla said they're going to sell THE EV for the masses and fund that endeavor by first selling EV sports cars to the rich. Well, they've eaten up a massive amount of masses' tax money to sell EVs to the rich and there's still not EV for those masses.
SolarCity, funny enough, is being bought by Tesla because it has failed to be financially solvent.
And the Hyperloop is 100% pure science fiction. And bad science fiction at that.
All promises, no delivery, and still praised for "vision" and "innovation".
This is what we should really be thinking about. We shouldn't design robots to work in the spaces we designed humans to work in. We should design the spaces with robots in mind. Take building a house. Currently a huge undertaking that requires a lot of manpower. In earnest though a house is quite a simple thing compared to many of the things that are assembled by robots today. There's no reason we couldn't change the construction process and have entirely robot built homes. Sure, it will affect jobs and the economy, but it's a thing that could be changed today. There are plenty of other things robots could do before we build a general purpose robot. Medical diagnosis would be better handled by expert systems, there are many more such tasks.