Forging the human–machine alliance

By Stefan Moritz and Kate Smaje

The current volatile and uncertain environment—with labor shortages, inflation, and a potential recession—has once again made costs, efficiency, and resilience top priorities for executives. Automation offers organizations a way to make progress across the board.

But reaping these benefits requires executives to reimagine their operating model and processes to truly integrate machines. Companies should take an expansive view of how to apply this technology. Machines1 can now do much more than weld car parts. They can learn how to sort fruit or predict when key factory equipment needs servicing. They can also manage customer service, distinguish skin cancer from millions of other blemishes, compete in (and win) complex games, engage in debate, and categorize thousands of legal documents.

The transformative potential of machines produces equal parts awe and concern. Too many workers believe machines are coming to take their jobs. Their fear often stems from a feeling of a lack of control and a fear of the unknown—the inability to see inside the black box. The problem with this defensive mindset is that it blocks us from seizing opportunities. It also means the future seems to be happening to us, with no chance for us to play a role in shaping it into something positive.

Garry Kasparov’s defeat by the IBM chess computer Deep Blue in 1997 is cited as a turning point in the human-versus-machine saga. But the typical narrow framing of the match tells only half the story. Kasparov did not abandon chess in a sulk. Instead, he helped invent centaur, or freestyle, chess. In this thriving version of the game, players can use computers during games, tapping millions of games played by machines to augment human decision making. This combination of human intuition and creativity with the overwhelming calculating power of computers creates a daunting competitor. Experts agree that, at least for now, human–computer teams play better chess than computers (or humans) alone.2

This finding underscores the need to take a fresh perspective: What if we really prioritize creating a future of work in which machines join the team instead of replacing us?

Now is the time to control that change by optimizing teams composed of both machines and people. When thinking of machines as partners instead of servants or tools, companies have the opportunity to supercharge their innovation and performance by building teams that augment human abilities rather than replace humans. Leaders will also be forced to reconsider what inclusion means in the context of human–machine hybrid intelligence and how it can be harnessed to solve more complex problems.

Companies that take these factors into account and move first to shape the future of work will have an easier time attracting the talent they need to implement new ways of working. The result can be a flywheel of innovation fueled by tech-empowered humans.

Tracking the partnership of human and machine

Machines are already having a massive impact on traditional tasks. Indeed, examples abound of machines working with humans, leading to the tasks being performed better together than either could do separately. Although about half of the tasks people perform today can be automated, only 5 percent of jobs can be fully automated.

Take medical imaging. AI image-classification systems can outperform human doctors at spotting cancers and other pathologies by being trained on millions of images. Despite the machine’s quantitatively better performance, though, doctors are in no danger of losing their jobs. A machine will not be trusted to make a diagnosis by itself, let alone deliver that news to a patient. Rather, by growing to trust the machine for rigorous diagnostic and research support, doctors can spend more time designing treatment regimens and nurturing the doctor–patient relationship. Fostering this human–machine collaboration can result in significantly better healthcare outcomes.

Human–machine teams also have the upper hand when it comes to the automation of production logistics, according to an interdisciplinary research team from several universities.3 Researchers assigned transport tasks to a human team, a machine team, and a mixed team. The tasks simulated using vehicles such as forklifts to make deliveries of production materials for an auto plant. The human–machine team was more coordinated and efficient and had the fewest accidents; it was, to the researchers’ surprise, the clear winner.

“There will also be many scenarios and uses in the future where mixed teams of robots and humans are superior to entirely robotic machine systems,” said Professor Matthias Klumpp of the University of Göttingen about his study on human–machine cooperation. “At the least, excessive fears of dramatic job losses are not justified from our point of view.”4

Taking a human approach to machine team members

Modern machine superpowers—fast and accurate computation and the ability to ingest terabytes of data—seem to be almost the opposite of some of today’s most sought-after human qualities: creativity, empathy, critical thinking, and emotional intelligence. Companies that design and plan for machine and human qualities to become complementary, rather than oppositional, will have the most effective teams.

The idea that diverse teams perform better than homogeneous ones should extend to include people and machines. We believe organizations can adopt the twin goals of creating an intellectual division of labor that distributes processing power and then building a culture that incorporates a collaborative, trustworthy hybrid intelligence.

Creating a culture of collaboration

When considering the intersection of machines and humans and how to establish a supportive environment, leaders should seek a deeper understanding of the end user’s needs and goals as well as a more holistic view of the opportunities available through interaction. A focus on five actions can help:

  • Redesign work environments for teams inhabited by people, machines, and data.
  • Rethink the work rituals and norms we live by in our roles to foster inclusivity and trust among machines and people.
  • Help machines interpret human movement, mood, and thought and respond with appropriate information or feedback.
  • Provide training and encourage experimentation for leaders and teams to learn, demystify, and create shared experiences to build trust.
  • Identify new opportunities for human–machine interaction and orchestrate pilots or lighthouses in a setting where solutions can be improved iteratively and then scaled.

Machines could also be programmed to do their share. Machines could be programmed to surpass humans at recognizing emotions and empathy, but granting machines the ability to act on information in ways that connect with humans will likely prove more difficult. If companies could build teammate algorithms that cause machines to behave in more inclusive and collaborative ways, machines could make humans far more enthusiastic about their jobs. This task is not easy, and the aim is not to make machines human. They are limited by what we enable them to do (as Janelle Shane’s recent book, You Look Like a Thing and I Love You [Voracious, 2019], demonstrates well), but we haven’t explored the hybrid intelligence well enough yet.

By extending the emphasis on collaboration and inclusion to machines, organizations may spark new thinking—and debate—about these issues and how to make them influence positive outcomes.

Communication is going to be a barrier for human–machine teams. But, as with all relationships, organizations must first put in the time to become comfortable with viewing machine intelligence as a peer rather than as merely a tool, and then determine the contours of that relationship.

The biggest opportunity for human–machine collaborations might be the potential of outlearning competitors. The human–machine debate is challenging what we know about technology and interactions: what a successful team might look like, how people and technology can interact for success, and what it means to be social. If an organization were to shift its approach to machines from “them versus us” to “we,” it could, in the most productive way, facilitate the continued integration of machines into the workforce and yield tremendous effectiveness and increased well-being.

Stefan Moritz is a senior design director in McKinsey’s Stockholm office, and Kate Smaje is a senior partner in the London office.

The authors wish to thank Rakhi Rajani, an alumna of the London office, for her contributions to this article.

1 For our purposes, the definition of machines includes machine learning and artificial intelligence—basically, any algorithm or technology that can help people work more effectively.
2 Mike Cassidy, “Centaur chess shows power of teaming human and machine,” HuffPost, December 30, 2014.
3 “Better together: Human and robot co-workers,” press release, University of Göttingen, May 24, 2019.
4 Ibid.