Robot descendants of Elon Musk are too stupid

Robot descendants of Elon Musk are too stupid


Things rarely fall flat for Elon Musk, but when the great showman last week hinted that Tesla will finally unveil a humanoid robot, the response was surprising.

Standing 5 foot 8 inches tall, Tesla’s Optimus looks like C-3PO from Star Wars redesigned in the sleek, faceless style of Sir Jony Ive. Musk reckons his latest offspring will revolutionise the economy. But for once the adulation that normally greets his new ventures was absent.

“Anybody who thinks Tesla is actually building a humanoid robot is living in an alternate reality,” wrote robotics pioneer Filip Piekniewski on Twitter. “There is absolutely zero credibility in these claims.” One analyst grumbled to Reuters that: “it’s all part of distracting people and giving them the next shiny object to chase after”.

The distraction jibe will sting: Tesla cars are struggling with automation in a very public way, with accidents alarmingly common. Having predicted full self driving by 2020 a decade ago, the whole sector has fallen far short, and is undergoing layoffs as the utopian vision recedes ever further into the distance. Tesla is no exception.

This reality-check should be a warning – tempering our expectations of dramatic robotic breakthroughs. It casts doubt on the proposition that a “Fourth Industrial Revolution” is about to break over us. In recent years, this grand and slippery phrase has been evoked by figures as diverse as Jeremy Corbyn and World Economic Forum founder Klaus Schwab, who devoted a slim volume to the subject in 2015. “Dramatic change is all around us and it’s happening at exponential speed,” Schwab asserted. But it is it?

All talk

Name-checking the Fourth Industrial Revolution has become a fashion statement: it places one firmly amongst the “smart” people who “are in the know”. The mantra is evoked to refer to many things – from fake meat to gene editing – but its defining assumption is that developments in artificial intelligence will revolutionise industrial robotics – leading to an era-defining impact on capital investment and the labour market.

This is the criteria by which we should judge whether it is happening at all. Awkwardly, however, this is a Revolution that is stubbornly refusing to appear. So far, it’s all talk.

Annually, automation installations have increased by around 6pc since the financial crash, the International Federation of Robotics reckons – a steady and very incremental increase. Globally, China accounts for over half of all the industrial robots installed, but this number is skewed by a glut of new electric car manufacturers opening their first plant.

In Chinese car factories, installations jumped by 89pc in 2021. However, automotive is traditionally a boom or bust market for robots, one that sees splurges of big investment as a new platform is produced, and then nothing. China’s credit-fuelled electric car boom may prove to be short lived. In retail and hospitality, the picture is less promising.

Amazon recently put the expansion of its cashier-less robo-stores on hold in the UK, while 236,000 Tesco customers have signed up a petition to halt the trend. Not all can be dismissed as Luddites – the robo shopping experience is often inconvenient. The tech just isn’t up to snuff.

UK plc could certainly be doing better. We’ve fallen to 24th on the league table of nations for investment in industrial robots. George Thompson, chairman of BARA, the UK’s trade association, recalls being astonished when visiting a factory that’s part of a major car manufacturer’s supply chain only decade ago, and proudly being shown a century-old stamping machine that was used to bash out parts for Spitfires in World War Two.

“That’s staggering,” he says. “In Germany or Denmark they’re keen to show you their spangly, new and very efficient machinery”. This is the sort of business investment Britain so desperately needs to boost productivity.

Demographic dearth

Keep in mind that economic conditions since Covid favour renewed investment in machines. Human labour has become more expensive, and Western countries, Japan and China are all facing a demographic dearth of young people willing to do lower paid manual work, such as packing or taping boxes, which can be easily automated using existing technology. With both factors in play, we should be seeing a much more dramatic spike in investment.

But on the technological side what we’re seeing is an annual incremental improvement, not a revolutionary one. It’s a pattern that’s consistent with the components of what’s called the “Third” industrial revolution, such as 4G data networks and cheaper sensors, being more widely deployed. However, the artificial intelligence magic required to create any sort of “Fourth” Revolution is becoming glaringly absent.

This may be surprising if you acquire your knowledge from the pundit class or LinkedIn gurus, where the Fourth Industrial Revolution is kept alive through webinars and earnest discussion sessions.

The fatal flaw underpinning the fantasy of the Fourth Industrial Revolution is the assumption that machine learning, the software technique on which today’s “artificial intelligence” is based, would translate to better real world robots.

Here we find a dilemma, known as Moravec’s Paradox, and robotics experts like Pienkewski have been warning us about it for years. For computers, the low hanging fruit of deduction is easy, and an AI can fool us by winning a game with simple rules – like chess – or by creating a visual pastiche. Yet computers find emulating the motor skills of a toddler incredibly difficult.

The mechanics of robotics are now very good, but robotics trying to use today’s artificial intelligence is really struggling.

On Friday, where Optimus makes its debut, don’t be surprised to see it develop stage fright. Be even less surprised if it doesn’t do anything useful at all.

News source

Leave a Reply

Your email address will not be published. Required fields are marked *

Pin It on Pinterest