The AI revolution has arrived. And although the technology is still in its infancy, it promises to radically transform the global economy—impacting human lives, culture, and politics in ways that we can scarcely imagine. A recent article by Forbes Technology Council member Christian Pedersen argued that artificial intelligence will create new opportunities for data scientists, researchers, analysts, and other highly educated technical specialists even as the more easily-automated job functions of low skill workers “fall to the wayside.”
Pedersen’s argument is correct on both counts, but the AI economy is also dependent upon one more ingredient: subject-matter expertise. Participation in the burgeoning AI economy doesn’t require an advanced degree in data science or fluency in the latest programming languages. In fact, many of today’s workers already have the sort of invaluable subject matter expertise that will be essential to helping AI become more sophisticated, efficient, and useful in the years to come. The future of industry will not be one in which humans are replaced by AI, but rather one in which humans and AI work together. That’s because, as it turns out, AI isn’t all that smart without us humans.
How to Train Artificial Intelligence
For decades, we’ve watched as advances in industrial automation have precipitated the gradual decline of low skill job opportunities in manufacturing and other blue collar industries. Now we’re beginning to see that AI has the capacity to replace skilled, white-collar knowledge workers as well. While it’s true that workers must evolve to stay relevant in a job market that will be increasingly shaped by AI and automation, this job market won’t just need technical experts with advanced degrees. AI-powered solutions require much more than just developers, researchers, and analysts to operate effectively. AI needs training.
For example, Volkswagen’s Innovation and Engineering Center California (IECC) recently debuted a new version of the classic 1962 VW Microbus, which was created by AI using generative design. The VW Type 20 Concept features bright orange wheels, side mirror supports, and other components that replace the usual straight lines and bulky forms of traditional car design with unusual, 3D-printed support lattices resembling tree branches or vines. These structures may look strange, but they provide the same level of support as standard components while using significantly less material.
Does this mean that generative design will eventually replace the human designers of the future? Not quite. Like most AI applications, generative design relies on human input to set constraints and handle complex decision making. As IECC Principal Product Designer Erik Glaser notes, “their software figures out a bunch of optimized results—some of which look insane—and then you pick the ones you like.” Even as AI evolves to take on high-level duties like product design, it requires human assistance to create something that humans will actually use.
AI designed for simple tasks like image recognition can be trained on data alone, but AI assigned to more specialized tasks needs input from human experts. Post-doctoral researchers, data scientists, and other AI specialists may be brilliant people with diverse skill sets, but rarely do they have insight into the nuances of product design, the delicacies of customer service, or the subtleties of concierge hospitality. A computer programmer can create AI capable of performing the most general tasks—answering a customer inquiry or checking in a hotel guest, for example—but only the knowledge and experience of subject-matter experts can help virtual agents become truly useful and effective.
Humanity’s New Partners
The workers of today can easily become the AI trainers of tomorrow, leveraging years of hard-won expertise to help AI achieve its full potential. This is not a far-flung hypothetical — there are already businesses (LinkedIn, Kixeye, and Nextdoor to name a few) implementing AI platforms that use non-technical experts for exactly this kind of work. These aren’t “one-and-done” engagements, with corporate AI agents consuming expert knowledge and discarding the humans behind that knowledge once they’ve outgrown their usefulness. The world is not static. As businesses grow, change, and introduce new products, their AI needs continuous training and improvement to keep up with the latest developments.
AI will transform the global economy and its workforce in profound ways, but it has the potential to be both a creative force and a destructive one. Businesses and policymakers must take an active role in ensuring that there is a place for the workers of today in the AI economy of tomorrow. This means steering businesses to help their workers understand the value of their expertise, and creating programs that will allow them to further develop and refine that expertise. It also means creating systems of royalties and residuals that will provide long-term compensation structures that reflect the enduring value of experts’ contributions.
Fortunately, AI has a genuine need for the expertise that many of today’s workers possess. A prime example comes from Garry Kasparov, the former chess champion who famously faced off against an early chess supercomputer. In the late 90s, Kasparov invented a new form of gameplay called “advanced chess,” in which human-AI teams face off against one another. Ultimately, he found that neither human nor machine intelligence could compete with the combination of the two, asserting in a 2017 Financial Times interview that “a human plus machine will always beat a super machine.”
Even thinkers who are far more pessimistic about the future of human-AI collaboration are forced to acknowledge the importance of cooperative interaction between humans and machines. For example, in his 2007 book, Super Crunchers, famed lawyer, economist and Yale professor Ian Ayres argues that humans assisted by algorithms will always underperform algorithms operating alone. This is a dubious assertion, but even Ayres must acknowledge that human knowledge and intuition is vital to setting the parameters in which AI operates. As he points out, “without theory or intuition, there is a literal infinity of possible causes for any effect.” He goes on to argue that “the hunches of human beings are still crucial in deciding what to test and what not to test.”
The truth is that AI cannot reach its full potential by working alone, and neither can we. Both humans and AI possess abilities that the other can augment, as well as limitations for which the other can compensate. That’s why, as we barrel inexorably toward our AI-powered future, we must ensure that the story of AI is not a story about humans being replaced in the workforce. Instead, let’s make it a story about how we humans found our newest and greatest business partners.
Antony Brydon is CEO and co-founder of Directly, an automation platform provider.