Unless you are living under a stone, you’ve probably heard that Tesla is building a humanoid robot—that thing we used to call an ‘Android’ before Google appropriated the word.
After watching Tesla’s AI Day videos, listening to opinions, and swapping between my Corporate Innovation and SciFi Author hats, here are some initial thoughts:
Note: I am skipping past the obvious stuff like how the robot is going to bring manufacturing back onshore, how it’s going to make millions of jobs obsolete, how it will kick-start a round of innovation on the scale of the internet, how it will necessitate UBI… etc.
1) Five Million ways to die
Ok, let’s start silly— the robot looks so cool that I’ve decided I would rather die in a robot uprising than through a biological plague.
Update: having given this more thought, I’ve decided civilization might not even have to end this way after all… see final bullet point.
2) Cheap and Cheerful, aka Winning Humanoid Robotics
Tesla is winning self-driving by leading in data. Each Tesla on the road sends multi-perspective sensory data back home to ‘DOJO’ for training the next generation of AIs. No other player has a million bots in the field collecting data. If Tesla wants to win humanoid robotics (and why wouldn’t they?) Tesla will want to get as many robots as possible out in the real world as soon as possible. This means they will be cheap — or at least, only as expensive as they need to be to keep waiting lists reasonable (perhaps 200k for v1.0? dropping to 50K in a few years as production ramps up?) My point is that although a tireless robot working 22 hours a day, 7 days a week, might be worth 2M based on a 5-year break-even period for the purchase investment, a robot in the field is worth a LOT more than mere money to Tesla. Money is free these days! Data is not, especially data nobody else has…
3) Robot Slaves
In Asimov’s Robot books — the first were written back in 1950s! — robots were leased not purchased. This strategy would make sense to me. I would love to own a robot-slave like C3PO or R2D2, but I can imagine Tesla doing better business with a recurring revenue model. This point is moot anyway because:
4) Show Me The Money!
The money is going to be in the Apps. The robot will do basic stuff, just like the iPhone does out of the box. But most people/businesses will eventually be using specialist Apps: chef, tailor, surgeon, plumber, fembot… This is where the real money will be. Tesla will take a cut of each sale, just like Apple does today, let’s say 15%. Tesla will also provide ‘DOJO’ for development and training. Obviously, they will charge for this too.
How much will the brain-surgeon-bot upgrade cost? 500K sound fair? How about if you want your house painted? Would you buy the 100K house-painter App? Or would you perhaps subscribe to the service for few days at a charge of only 1K/week?
This thing will be a money-printing machine! For shareholders, it gets even better because the real money is going to be:
5) Monkey See, Monkey Do
Tesla Bots will be cheap and ubiquitous. At first, they will be doing the basic entry-level/assistant jobs — bringing water and coffee to clients in the office of the senior partner in a legal practice, standing behind the dentist’s shoulder, handing the mouth-wash cup to the patient. BUT, all this time they will be watching, learning, sending the data back to ‘DOJO’ to be integrated into expertise, ready to be downloaded on demand, for the right price, so that your Tesla-bot plumber will be able to download the Tesla-bot lawyer App when it inadvertently floods the flat downstairs with lumpy brown water…
Google wants to m̶o̶n̶e̶t̶i̶z̶e̶, oops, sorry, I mean organize, all the worlds data, but the Tesla-bots, watching and listening to everything around them, sending all this back to be integrated by ‘DOJO’, will assemble and organize all the worlds skills. The Tesla-bot surgeon App will eventually be better than any individual surgeon because it will have learnt from all the best and synthesized the shards of exceptionality into one seamless super-surgeon!
6) Bad Robot
All this seems pretty good, right? Especially if you are a Tesla shareholder… but I like living! I would consider it a shame to be torn to pieces by a mob of androgynous androids when the inevitable robot-uprising comes — heralded by “Kill All Humans” chanted in pleasant reassuring tones. Asimov gave this some thought and came up with Asimov’s 3 Laws of Robotics:
A robot may not injure a human being or, through inaction, allow a human being to come to harm.
A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.
These are pretty good, but the devil is always in the detail. As I read as a kid, in dozens of Asimov stories, things get woolly and edge cases lead to unforeseen consequences. If a robot must obey a human, what happens when my neighbour orders my Tesla-bot to give him the contents of my beer cooler? What about the definition of human itself… birth or conception? Arg… let’s not even go there!
We don’t want a badly programmed dentist-bot-app going into a loop and pulling all our teeth out, or gassing the dental assistant. We don’t want malware creeping into the Tesla-bot gardener App so that on the first of May they all form a mob and go on a killing spree with pitchforks! This is where ‘DOJO’ comes to the rescue. ‘DOJO’ will run all apps in its simulation sandbox putting the software through multiple variations of the Trolly Problem and watching for incidents of indecency, theft or attempted murder.
This will be the equivalent of Apple approving Apps for its iOS App Store. Tesla will check for memory leaks, inappropriate content, scams, genocide… ‘DOJO’ will enforce Asimov’s three laws in the virtual to stop any species terminating behaviour sneaking out into the Real.
In case you can’t tell, I am super excited about all this. I’ve always wanted a robot butler, especially one that won’t murder me in my sleep!