When machines move on without us
[Published 10.19.2016]
Human labor may decouple from global supply chains but it won’t abandon regional and local productivity.
“You're thinking of this as you'll take an existing product and add some AI to it. That’s not what we’re seeing. What we’re seeing is an entirely new kind of product that wasn't possible before.” - Marc Andreessen
“Just as mechanical muscles made human labor less in demand, so are mechanical minds making human brain labor less in demand.” - Humans Need Not Apply
There’s a long-wave macro trend that’s been removing as many inefficiencies as possible from large-scale economic structures like supply chains, logistics platforms, and banking networks. For half a century, we’ve mostly applied crude-but-effective automation but now optimization is starting to amplify massively as globalized value chains and their ginormous data sets become the playgrounds of neural networks and learning systems.
It’s an exciting and hopeful prospect to imagine that our brute-force civilization building might suddenly soften its step on the planet and its denizens, led by cognitive systems capable of grasping the insane amounts of complexity we’ve wrought upon the world. But this brand of progress will also fundamentally reshape labor, shifting humans out of the largest and most productive employers.
Nature always seeks to optimize, for fitness in a competitive landscape and for efficiency in energy use. Google has used AI to draw down cooling in data centers by up to 20%. Imagine the potential to manage energy and carbon if this could be used on all industrial systems. Researchers at Stanford can predict poverty by feeding satellite images of the Earth to AI classifiers. They can see patterns unnoticed by humans, offering hope for remediation of economic suffering. Rotterdam’s shipping port (Europe’s largest) now relies on industrial robotics and machine intelligence to manage its role in global logistics. The complexity of our supply chains now defy complete human understanding and management. And there’s no shortage of programs training autonomous vehicles to drive better, more efficiently, and more safely than humans. Machines can manage fuel consumption better than us leadfoots, and they react much more quickly to the unfolding present.
AI is different. Previous technological revolutions created more new jobs than those that were eliminated. But now our tools are taking over non-routine jobs. As a labor class, autonomous systems are radically more competitive with humans than previous technologies. As others have noted, if your work is routine, repeatable, and/or data-intensive, AI will soon do it better and for less money.
By cost, capability, and efficiency, our tools are entering an exponential at a time when human productivity has mostly flattened. Humans, like the horse in 1915, may be looking costly and inefficient.
Foxconn replaced 60k factory workers with robots. Amazon bought Kiva Robotics to deploy 30,000 robots in their warehouses, cutting costs by 20% and reducing the need for parking space around the buildings. McDonald’s and Wendy’s eliminated low-wage employees for self-service kiosks and mobile apps. In banking & finance, Germany’s Commerzbank will automate 80% of its functions by 2020, eliminating 9600 jobs. Dutch lender ING is replacing 5800 jobs with automation and autonomous systems.
It won’t just be the most well-heeled corporations and governments deploying AI for efficiency. Barriers to entry are falling as these tools get cheaper and easier to use, bundled into cloud platforms like AWS, Azure, and Google Cloud. Now, AIs have APIs. Like the computer and the Internet, most will use learning systems. They will shape our institution and our behaviors.
The World Economic Forum anticipates a striking landscape in 2020, just 3 years away. Global employment is forecast to rise 1.73% - a rather dismal “end of growth” number but in the study it's buoyed by a clutch of emerging technologies referred to as the Fourth Industrial Revolution. The largest job contributors cited are in some of the most structurally-difficult and intractable areas: young demographics in emerging markets (ie high risk, frontier), and in the opportunity to move more women into the workplace. So, not a lot of easy upside. Indeed, unpacking the WEF report shows a net loss of 5 million jobs by 2020.
We humans see AI as driver of negative job growth but the largest employers on the planet see it as an engine of performance and profitability.
A report by Bank of America Merrill Lynch says "we are facing a paradigm shift which will change the way we live and work. " They continue: "We anticipate... the possible displacement of human labour (with 47% of US jobs having the potential to be automated) and growth in inequality (with a 10% supply and demand gap between skilled and non-skilled workers by 2020).”
Western labor productivity has slowed dramatically since 2000. Humans can only do so much. And demographics are skewing older. A 2015 RAND study anticipates “a potential surplus of low-skilled workers around the world and a global shortage of medium- and high-skilled workers.” Low-skilled labor is what gets displaced by automation and the bar for “low-skilled” is rising as automation evolves into AI.
It’s not just low-skilled labor that will feel the impact. It’s any work that requires a high-volume of information processing or management of highly-complex systems. If learning systems continue to accelerate as rapidly as they have in the past several years, they’ll remove human jobs faster than we can replace them. AI will become the new offshoring.
BofA Merrill Lynch also forecast a 50% chance of generalized AI by 2040-50 and a 90% chance by 2075. That’s AI that can do anything we can do. That’s when we’ll need a guaranteed basic income. Or full-on luxury communism.
When the top of the economy organizes around machines, humans organize around each other. Wielding AI to augment and amplify human-scale networks, we respond with urban collectives, rural farming communes, and local manufacturing, all tapped into information systems and global networks. Profits accruing to the top and intense competition in urban centers reinforces the Darwinian flavor of late-stage Capitalism. DIY collaboration is an imperative, and it’s a long journey between autonomous systems running deep-bore mining sites at the edges of civilization and Joe Sixpack at the Apple Store. It’s a complex web of M2M and B2B extracting and shaping value and burning tons of energy along the way. We’d expect a living system like that to optimize for shorter distances.
Brooklyn’s New Lab occupies and remediates the husk of an industry that long ago abandoned them. As a public-private partnership, the New Lab re-claims the Brooklyn Navy Yard as a hub for the Fourth Industrial Revolution - robotics, local manufacturing, and intelligent soft systems. They’ve created a shared social space, aiming to bootstrap innovators into companies that will hopefully scale to become significant job providers - or at least become part of a web-work of small-scale producers serving dense urban centers. The world is rapidly moving into cities and much of the next 50 years will be about organizing and optimizing for urbanity. Energy efficiency insists that we become more regional and local.
Gotham Greens is the world’s largest rooftop farm. Now they have several of them in New York and Chicago. Their newest in New York will farm 60,000 sq ft for a yield of over 5 million heads of leafy greens annually. You can deliver the goods by lorry or bike, serving the 8.5 million residents of New York City. This is when over-extended and hyper-optimizing global supply chains start to retract and find more energetically efficient models. This is when climate and nature - The Thing We Live Inside Of - start radically pricing in to the global economy.
These are just two examples of many emerging in response to numerous systemic conditions, and empowered by unprecedented access to information, collaboration, and technology. Human labor may decouple from global supply chains but it won’t abandon regional and local productivity.
The nature of technology is to get cheaper and more powerful and into more and more hands. We’ve wrapped supply and manufacturing chains around the planet to deliver easy access to just about any material, component, or assembly. We have a global garage of parts and tools and knowledge and it’s really firing up right when a ton of deferred costs are coming due and there’s a weird digital deus ex machina stepping onto the stage.
With incredible visibility into the contemporary human condition, the imperative is to understand what makes us unique, how we differentiate from machines and learning systems, and how we create enduring value in human-scale economies while maintaining and optimizing the global systems we all depend on.