By Adam Dorr
In my previous blog post, we explored some of the remarkable and non-obvious implications of humans being able to remotely operate humanoid robots via immersive virtual reality interfaces. The overarching takeaway message was that we can think of this as a kind of teleportation, allowing someone to instantly relocate their mind and some of their physical agency to a distant location.
But it isn’t just humans who will make use of this sort of teleportation. AI agents will be able to move in and out of robotic bodies far more easily than we will, and that capability is a game-changer.
Always at Your Side
Perhaps the most immediate example of AI teleportation is an extension of the idea that your “settings” following you wherever you go: your personal AI assistant won’t need to live in a single device that must physically accompany you as you move, but instead will be able to occupy whatever available devices happen to be near you at any given moment. Two obvious and very promising examples are autonomous vehicles (AVs) and humanoid robots.
Today, when you ride in a Tesla robotaxi, your settings already follow you: as soon as you enter the vehicle, your music and other preferences take effect. Within a few short years, your personal AI assistant will be able to do the same with humanoid robots. There are some subtleties here worth unpacking.
In a robotaxi, there might not be such a huge amount of value in having your personal AI assistant’s personality take over the vehicle itself. Beyond the relatively minor benefit of using the robotaxi’s speakers for a nicer sound experience, your personal AI might as well just remain as a presence on your phone. But with humanoid robots, things would be very different.
Imagine it’s 2035 and you're out shopping at the grocery store with two young children in tow. Your personal AI assistant with months of training operating your humanoid robot at home would know far better than a standard retail store’s bot AI how you preferred to interact with it, how best to help you wrangle the kids, where you need the close support of a helping hand versus where you prefer space, and all of the other nuances that are exactly akin to training a human personal assistant. And on top of that, we humans are virtually certain to become attached to our personal AIs emotionally. So, for both practical and psychological reasons, many if not most of us will prefer the familiarity of our own personal AI operating the grocery store’s humanoid robots than a standard retail AI. And like the Tesla robotaxi that allows you to seamlessly bring your preferences into their robots-on-wheels, other businesses will naturally allow you to do the same with their robots-on-legs.
Expect platform operators to capitalize on network effects around these sorts of beneficial services: “take your AI with you anywhere you go, using our fleet of droids!”
Ubiquitous Expertise
Commercial and government AIs will also be able to teleport into any robotic platform on demand. That means any humanoid robot can instantly become any kind of expert. Because the cognitive and physical capability – the mind and body – of robots are not inherently coupled, ideas like training, specialization, and roles will cease to be meaningful. A humanoid robot can be a babysitter in the morning, a farmhand at lunchtime, a firefighter in the afternoon, a neurosurgeon in the evening, and a warehouse worker all night. Moreover, it makes little practical difference whether these capabilities are delivered by a single superintelligent AI that is expert across all domains, or through collection of smaller, more specialized AIs. The result will be the same: expertise available to everyone, everywhere, all the time.
It is also worth noting that a superintelligent AI whose mind is spread across multiple robotic bodies might be able to utilize that embodiment for capabilities that are difficult for us to imagine today. What would it be like to have a “body” consisting of five humanoid robots? What could you do with 10 arms? (An elephant might ask us a similar question – “what would you do with two trunks?”). And what about a body consisting of hundreds or thousands of robots? Science fiction has, of course, explored such questions before. But now the reality is imminent.
The Aliens Have Landed
For the entire 600-million-year history of animal life on Earth, mind and body have been inseparable. But now AI and robots that are able to decouple software from hardware are here. This is a deeply strange and totally novel condition. It really is something new under the sun. We will readily adapt to the change, as we always do. But given past experience with radically transformative technologies, it would be prudent to expect the unexpected and proceed with caution. There are sure to be unexpected consequences, both for better and worse, that come from the separation of mind and body. It’s time to start thinking seriously about them.
This is the second blog in a series on teleoperation. Read the first blog here.