Analysis: Tesla’s humanoid robot might be Elon’s dumbest idea yet

Let’s just dive right in. Tesla held its much-ballyhooed “AI Day” yesterday which, as Shift’s Cate Lawrence pointed out earlier, was little more than a job fair.

There were, however, a couple of interesting reveals.

The good news was the company’s new AI chips. According to the supposed specifications and obligatory hyperbole, Tesla’s AI chips will be the world’s most powerful.

It’s at least plausible that the AI team at Tesla could pull off such a feat. The race to develop a better AI chip – currently dominated by Intel, AMD, NVIDIA, and the like – is a perpetual game of leapfrog.

But let’s talk about the Tesla robot.

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!

TNW’s Ivan Mehta covered the announcement and all the machine’s specifications earlier today.

So, for this analysis, I’m going to give you my conclusion first, then we’ll work our way back from there.

Closing thoughts: Everybody wants this to be real. Me, you, the entire writing staff of the Simpsons (wait for it, I’m sure it’s coming), even Tesla’s competitors.

We all want Rosie the Robot to be real.

But here’s the truth laid bare: this is a hustle. The Tesla Robot is Elon Musk at his PT Barnum-esque best. He’s promising everything you want and daring you to dream along side him while he picks your pocket.

Okay, now that we’ve gotten that out of the way, allow me to explain why.

The kind of AI it would take to power a fully autonomous robot doesn’t exist. Rosie the aforementioned Robot, from the Jetsons, is not real. In fact, and in spite of my enthusiasm for the idea, I doubt we’ll see a fully-autonomous robot in the next 30 years.

Tesla claims that it’s going to build a robot with the same AI brain as its cars. The robot would operate on a vision system using multiple cameras. Nothing about that idea even comes close to the technology already in the space.

But let’s continue anyway.

Engineering-wise, it can carry about 2 or 3 sacks of potatoes, it moves about as fast as you can walk backwards, and if it was a UFC fighter it’d weigh in as a flyweight.

The obvious use-case for such a machine is as a factory worker, but even a cursory glance at the design rules this out almost instantly.

For crying out lout, it has humanesque hands and could be knocked over by a saloon door.

What’s it going to do, offer lemonade to the human workers and iced oil to the robots that were actually designed for factory work?

It’s obvious the Tesla Robot is meant to appeal to consumers. The design isn’t meant to be functional, it’s meant to imbue us with warm, fuzzy hope.

It’s just a lil’ bitty thing y’all, c’mon it can’t hurt anybody!

We’re supposed to look at it and imagine all the ways it could make our lives easier. According to Tesla’s website, it’ll only be capable of performing simple, boring, or unsafe tasks at first.

But the claims Musk makes about it’s future abilities are straight out of a Will Smith summer blockbuster.

Speaking at yesterday’s event, Musk said:

Can you talk to it and say, ‘please pick up that bolt and attach it to a car with that wrench,’ and it should be able to do that. ‘Please go to the store and get me the following groceries.’ That kind of thing. I think we can do that.

Okay, sure. Sure, okay.

That’s not how AI works. Just ask Boston Dynamics. While Tesla was busy contemplating the best mannequin design to use for its big ‘robot’ reveal, the makers of the incredibly popular Atlas and Spot robots have been making humanoid robots dance, perform parkour, and stun the world with backflips.

This is nothing. In a few years, that bot will move so fast you’ll need a strobe light to see it. Sweet dreams… https://t.co/0MYNixQXMw

— Elon Musk (@elonmusk) November 26, 2017

Here’s the rub: Boston Dynamics was founded in 1992 as an MIT spin-off. That puts Tesla about three decades behind the curve here.

But, just for the sake of argument, let’s pretend that Tesla has a robot as advanced as Atlas. There is a zero-percent chance that a Tesla robot running any AI system on the planet will be able to perform a function as simple as walking into my house and fixing me a cup of instant coffee.

I’m not even suggesting it do anything radical like carry my laundry up the stairs and put it away or – laughing out loud here – going to the grocery store and picking up a list of items.

There’s a huge difference between building a dedicated burger-flipping machine that’s bolted to the floor or a warehouse bot with wheels that follows a digital pathway and creating something that can navigate any building.

The sheer number of systems and hardware it would take for a completely autonomous machine to enter a novel space (like my house) and then locate the ingredients (instant coffee, water, my mug, a spoon) to make me a cup of coffee would be astronomical.

I’ve got a kid running around, toys everywhere, half the time we don’t put the coffee can back where it goes, my mug could literally be anywhere, and good luck finding a clean spoon!

(Related: Tesla misleads customers about self-driving features, senators allege in request for FTC probe, on CNBC)

Now, a person could easily handle this task. I can walk into your mansion, apartment, or house right now and figure out how to make a cup of coffee as long as you’ve got the right ingredients. But there isn’t a robot in the world that can do that safely without training on the environment previously.

And, sure, I suppose if you wanted to prove me wrong you could dedicate some AI development to a coffee-making robot that specialized in rummaging through cabinets.

But then, if I asked you to make it fix me a grilled-cheese sandwich you’d have to go completely back to the drawing board and design an entirely new AI system.

So what happens when the Tesla robot tries to do something it hasn’t been explicitly trained on? The exact same thing that happens when you ask Alexa or Siri to do something it can’t.

A humanoid machine designed to do menial tasks sounds like a really stupid idea. If it’s supposed to carry things, build it to look like a wheel-barrow. If it’s supposed to do dangerous jobs, design it for the actual job. This thing is only fit for work as a crash test dummy in its current iteration.

At the end of the day though, Tesla stocks are rising and that’s surely all that matters to Musk and his sycophantic supporters.

Musk makes promises. They don’t come true. He makes more promises.

If @Tesla releases a fully autonomous car, I'll join @kaifulee and @rodneyabrooks and eat it. Bumper to bumper. With @CholulaHotSauce.

— Tristan Greene ?‍? (@mrgreene1977) October 24, 2019

It’s always a safe gamble to assume Musk’s just blowing smoke up our butts. Just glance at the above tweet.

But that was yesterday’s BS. It’s a new day, Musk’s got a new hustle, and it’s time for a new bet.

So here’s my new challenge to Musk: If Tesla can send a robot to my house (site unseen) next year, that can perform the same chores I currently ask my 4-year-old to do every evening before bedtime, I’ll roll it up in a giant tortilla and eat it with ghost pepper hot sauce.