As automation and AI continue to transform businesses across the globe, the tech industry is in the process of building a world that will look very different from the one we know today. This implies profound changes in how business leaders will structure their companies and suggests a shift in the skills required for success. We must start preparing for these changes now.
We can disagree about the number of jobs automation will replace, but most experts who study this closely predict monumental shifts in how we work. McKinsey projects that as many as 800 million workers worldwide could lose their jobs to robots and automation by 2030 — equivalent to more than a fifth of today’s global labor force. A previous study from Oxford University concluded that almost half of all U.S. jobs will be “susceptible to computerization” in the next decade or two.
Automation has already transformed manual industries, along with routine tasks that involve simple, rule-based activities like sorting mail and bookkeeping. But the wave of technologies referred to as “AI” — machine learning, computer vision, and natural language processing — allow companies to hand over increasingly sophisticated tasks to machines. GE and Shell, for instance, both employ algorithms for managerial work. An example from Shell is the use of machine learning to match employees with the right projects for their skills.
As Stanford University academic Jerry Kaplan writes in his book Humans Need Not Apply, automation is “blind to the color of your collar.” Whether you’re a factory worker, a paralegal, or a sales manager, automation is coming for your job over the next 25 years.
Yes, AI could be smart enough to take your job
It seems unlikely that half of tomorrow’s workforce will be unemployed, as the Oxford study implies, and likely that automation will create new classes of work even as it destroys existing ones. But the skills needed in the future will be very different from those our education system selects for today. In the future, companies will highly reward skills that are not currently valued by this system, like creativity and emotional intelligence, as these are among the skills computers will find hardest to replicate.
Conversely, the bureaucratic and administrative skills that our education system is geared toward providing will be far less relevant in the market. In fact, it’s likely these jobs simply won’t exist.
To understand what future organizations will look like, it’s helpful to think of the popular mantra that every business will become a software business. What does that mean, really? It implies that every business will leverage software to the greatest extent possible to outcompete its peers by “outsourcing” most traditional business functions to companies that specialize in those areas. This will be done to reduce costs, build better products, and ultimately, generate more profit. The businesses that thrive will be those that use software most effectively and wherever they can.
Automation is a leading indicator of this transformation, and it has rapidly moved beyond the limitations of simple, repetitive tasks and into areas where a self-learning algorithm can make decisions with sufficient certainty. And it can do this at a speed and scale that humans simply can’t. Thus, legal teams use machine learning to sift through millions of documents to find those relevant to a case, sales teams use it to identify targets and upsell opportunities, and financial advisers use algorithms to provide investment advice. These changes are happening today, and it’s naive to think companies will not apply automation to ever more complex tasks in future.
Experts often note that computers still can’t come close to thinking like humans. Computer scientist Edsger Dijkstra offers an interesting counter-opinion by saying that asking whether machines can think like a human is “about as relevant as the question of whether submarines can swim.” If a computer can perform the same task better, the process it uses to get there makes little difference. “We are approaching the time when machines will be able to outperform humans at almost any task,” Moshe Vardi, a computer science professor at Rice University in Texas, has said.
If software becomes better at any task where data can be brought to bear, from optimizing supply chains to designing products, there are few roles left where human judgment will be a superior option. Some industries will likely always benefit from a human touch — the artisanal coffee shop, the hospital ward — but within the walls of business, software will perform more and more analytical, administrative, and bureaucratic functions.
Honing valuable human skills in an automated future
The enabling force is human-in-the-loop AI — where algorithms perform business functions and a human steps in when the software is uncertain of the answer. An operator can feed human judgment back into the algorithm so it learns to tackle the problem better in the future. Human-in-the-loop AI greatly increases the scope of work that AI can perform because it allows software to handle tasks that are traditionally considered too nuanced for a computer to deal with.
In this model, software can handle almost anything to do with logistics, operations, and objective decision-making. What it can’t do is the creative work. A human will be necessary to craft the marketing copy that strikes just the right chord with other humans. The writer can then feed the copy into a machine that A/B tests it across target audiences, refining and personalizing it along the way. In this scenario, the algorithm does the targeting, but a human creates the initial words that compel an emotional response.
This example implies organizations that will look vastly different from today. We’ll still need humans to set the company vision and developers, designers, and creatives to build and program the software. But an entire tier of employees will become redundant. Think about anyone you work alongside who performs their primary tasks using software — that’s a long list which includes sales, finance, HR, accounting, marketing, and office management functions. Instead of building applications that humans can use to do work, we’ll increasingly build applications that perform the work itself.
This raises important questions about policy. We need to have a conversation as a society about how we will deal with mass automation. Do we prioritize the freedom of corporations to maximize productivity and competitiveness at the expense of secure employment? Or do we enact policies that protect certain classes of jobs? That’s not as far-fetched as it may sound: The Stimulus Act of 2008, for instance, focused on large-scale infrastructure projects, prioritizing employment over productivity. On a local level, San Francisco politicians voted to limit delivery companies to three robots each, clearly a move to assuage voter fears of automation.
Our education system will also need to evolve. Today, schools churn out graduates optimized for the type of rote, administrative work computers are already more adept at handling. Few children are encouraged to pursue creative, interdisciplinary subjects or develop empathy and interpersonal skills, yet those are the attributes we will most need in order to augment computerized decision-making. Ironically, when I look around at other leaders in Silicon Valley, those skills are massively over-represented among my peers.
Building a deliberate future for human employment
Whatever we decide, we need to proceed deliberately and in a way that balances the imperatives of business with what’s best for society. As Satya Nadella noted at Microsoft’s recent Ignite conference, the tools we build must ultimately contribute to our wellbeing, not detract from it.
“How are we going to use technology to empower people?” Nadella asked. “Every piece of technology should help embellish the capability of human beings. We definitely want more productivity and efficiency, but we do not want to degrade humanity.”
Fred Stevens-Smith is the CEO of Rainforest, an on-demand QA solution that improves customer experience by enabling development teams to discover significantly more problems before code hits production.