Recently thecompany 1X announced NEO, an in-home autonomous (sort of) android all purpose robot, for a cost of $20,000. This has been a vision of futurists, tech enthusiasts, and sci-fi fans for decades. Who doesn’t want a Rosie the Robot to do all their domestic chores? But is the tech really ready?

The home is not a friendly environment for an autonomous robot. It is chaotic, may include children and pets, has a lot of breakable stuff, and is optimized for people, not robots. Still, my Roomba does fine – it is a smart autonomous robot that does one thing (vacuum) reasonably well for a reasonable price. But what about all my other chores? And if the home is built for people, then why not a humanoid robot? The question is – are we there yet?

I spoke with Christian Hubicki, a roboticist, about NEO (if you’re are a patron you can see this on the SGU livestream) and he was impressed with the price-point. Twenty grand for an android robot is pretty good. It seems to function well, in that it can walk and pick up things. It is designed to be soft and is powered by current chatbot level AI. A the very least, it’s a walking chatbot.

But can it do a range of domestic tasks as 1X claims? Veteran roboticists and robot enthusiasts know that there are basically three ways to get a robot to perform a task. They can be autonomous, using sensors to sense their environment combined with AI programming to interact with it and complete a task. Or they can pre-programmed to perform a specific precise sequence of movements. This method works well in controlled environments, like a factory floor, to perform very specific tasks. Or they can be teleoperated, with a human in the loop partly or completely controlling what the robot does.

So whenever you see a robot demonstration (and it’s not a guy in a robot suit) you need to ask – is it doing what it’s doing autonomously, pre-programmed, or teleoperated? For NEO the company claims it is designed to operate first autonomously. However, for more complex tasks it may need expert help, in which case you can schedule a teleoperator to take over and complete the task. The company website says:

“NEO works autonomously by default. For any chore it doesn’t know, you can schedule a 1X Expert to guide it, helping NEO learn while getting the job done.”

What jobs would those be? Early reports indicate that this is pretty much any domestic chore – so don’t expect it to put away your dishes, fold your laundry, or straighten up your kid’s room without teleoperation. Expert mode also requires a monthly subscription (Christian hears this will be $500 per month and you get three years included in the cost of the robot, but I don’t have a confirming reference on this). Their site says you can rent NEO, including the services, for $499 per month.

So is this, then, just an expensive domestic helper? Do you really want to let a random teleoperator into your home? Why not just hire an actual person at that point?

Also, Christian pointed out the “helping NEO learn” bit – they are likely trying to sell the robot early, before it is fully baked, because they want the training data. It is likely NEO will get better with this training, and perhaps it will be able to do more tasks over time, but it remains to be seen if it will get to the point that you would feel comfortable with it handling your china.

All things considered it does not seem we are there yet. NEO is for super early adopters who want the idea of a home robot, and it can be a sort-of companion, but it is not a cost-effective way to do these light chores and many people might be squeamish about letting teleoperators into their home. Having to schedule a time to have chores done is also an inconvenience and it remains to be seen how good this service will be.

I see the utility of having an all-purpose robot, and of having an android robot for the home. Part of the success of the iPhone was that it was so damn useful in so many ways. Walking around with a computer in your pocket opened up a world of possibilities. Having a truly effective and reliable autonomous robot in your home could be equally useful, but it does not seem like the tech is there yet (or else NEO would not need a teleoperator).

The question is – how long will it be before we get there? Will NEO or another company’s version be ready for prime time in a few years, or will it take a few decades? AI is advancing quickly, but already we may be running up against diminishing returns. Robots in the home need to be super reliable and safe (like self-driving cars) and getting those last few percentage points of reliability can be a real challenge. As I have pointed out before – even when technologies advance non-linearly, problems can also be non-linear, so it can still take a long time to solve them.

In the end such questions are often resolved by the end-user – will people feel that the benefit of a NEO is worth the expense? How will people use it? If people find lots of good uses for a robot with this level of capability, and companies build on this by adding more clever functionality, we could be looking at the next iPhone. Or perhaps android robots will have use other than what the company is anticipating (like the microwave, which people do not use for cooking, but became and indispensable heating tool). Or perhaps it will be like the Segue, which worked wonderfully, but didn’t really find sufficient use. The fate of NEO and similar products may be difficult to predict for this reason – but the odds are it will fail unless companies can figure out how they are supposed to be used that makes sense and is worth the expense.

For example, the company claims that NEO will help you save time, but will it? For autonomous tasks it will likely need a lot of supervision, and you may want to secure some breakable valuables. If you have pets or young children, good luck. Lots of stairs is also an issue. While NEO is light for a robot, you don’t want a 66 lb robot tumbling down the stairs because your cat ran through its legs. Scheduling tasks and setting them up might be more trouble than it’s worth. If I have a dishwasher full of clean plates and glasses, am I really going to schedule a time for someone else to empty it? We also have to see how people react to the idea of the teleoperator. Is this a non-starter or will people come to accept it?

On the other hand, it may turn out that having a companion who can do simple but annoying tasks may be super helpful for some people. Such robots may find a niche, in elder care, for example, that bootstraps the products for a more general consumer.

We can say that for now, NEO is not the autonomous robot butler you have always wanted. It’s autonomous capabilities are limited, and there are very real concerns about its utility in the unstructured and chaotic environment of the home. Therefore, if it will succeed or flop depends on variables that are difficult to predict – once the expensive novelty wears off, will people find it useful or a hindrance: will they accept the downsides or not?

The other question prompted by NEO is this – is an all-purpose android robot the way to go for right now? We already have a model for a successful in-home robot – the Roomba. This is a task-specific non-android robot. Yes, that means it can only do one thing, but it works. Perhaps we should build on this model, developing in-home robots still designed for just one or a limited range of tasks. They also don’t have to be humanoid. Perhaps a four-legged dog-like robot is a better choice. Maybe a robot that can do the laundry and nothing else is a better trade-off – cheaper, safer, more reliable, and better at that one task (without needing to schedule a teleoperator).

The good news is – it seems likely that we are poised to find out the answer to many of these questions. I hope that NEO does not fail so hard that is sours the industry on the idea of in-home robots. I do think they are very likely inevitable. But what path we take to get there matters.