Is the Pandemic Causing Robots To Be Overrated?

Robots perform well in controlled environments doing repetitive, routine tasks.

That reflects neither the reality of the battle against Covid-19 nor the broader global economy. JL

Sameer Hasija and Aarti Gumaledar report in Harvard Business Review:

In the midst of the COVID-19 pandemic, robots deliver critical tasks in healthcare facilities and (are_ contagion-proof workers. Robots have even been extolled as “heroes” helping to “manage” the pandemic. Companies that have spent billions on adopting AI-powered robots report being disappointed. The current generation of robots perform routine tasks in controlled environments. If the environment is variable and unpredictable, robots will likely fail to perform. As Elon Musk famously pointed out, “humans are underrated”.

In the midst of the COVID-19 pandemic, robots have been having a moment. Media outlets across the world have reported on robots successfully delivering on critical tasks in healthcare facilities and their effectiveness as contagion-proof workers in many other settings. Robots have even been extolled as “heroes” helping to “manage” the pandemic.

It’s easy to see why. Robots are carrying out essential work such as disinfecting surfaces, checking on patients’ vitals, and delivering food and medicine to patients in quarantine. These solutions minimize human to human contact and reduce the risk of health workers contracting the infection from patients and surfaces in hospitals. The unprecedented labor shortages resulting from lockdowns around the world have encouraged even the most reluctant of users and businesses to adopt new technologies.

Unsurprisingly, robot manufacturers are seeing a surge in sales and interest. Robotics experts like UC Berkeley’s Ken Goldberg believe that this trend is likely to accelerate further and an article in the The Robot Report, a widely-read online magazine, predicts that a post-pandemic future will entail an even greater reliance on such automation technologies due to cost savings and productivity gains.

Should we believe the hype? For a company to maximize the benefit from switching to robots, two conditions need to be met.

1. The robots must work on bottleneck tasks.

Readers of a certain age will recall reading Eliyahu Goldratt’s 1984 management classic, The Goal, a fictionalized story of one manager’s professional journey. At one point in the book, the wise mentor Jonah points out to the book’s protagonist, Alex, that the latter’s plant has likely started underperforming because of the state-of-the-art robotics that Alex has just been proudly describing.

Jonah points out that unless the robots are used to replace bottleneck resources, there will be no gains in overall productivity, no matter how fast the robots work. And if no headcount reduction results from deploying the robots, there will be no significant cost savings. What’s more, if units operating robots utilize them continuously and show substantial efficiency gains in silos, then it is likely that the organization’s performance as a whole is adversely impacted by the robots.

Counterintuitive but true — both at Alex’s plant and in reality. It is why many companies that have spent billions on adopting AI-powered robotic solutions have reported being disappointed with their performance. Increasingly there’s evidence to suggest that, as tech pioneer Elon Musk famously pointed out two years ago, “humans are underrated”.

2. The task environment must be controllable.

In spite of recent technological leaps, the current generation of robots perform routine tasks in controlled environments. It is for this reason that retailers like Walmart and Giant limit their robots to simple tasks such as tracking inventory on aisles or spillage on floors.

If, however, the environment is highly variable and unpredictable, robots will likely fail to perform and many even be dangerous. Put simply, the driverless delivery vans used in locked-down Wuhan, where traffic was non-existent, will not necessarily work when the streets are busy again. Of course, this is not to say that delivery vans cannot work if the environment can be brought under control. If cities introduce dedicated pathways for autonomous delivery robots in densely populated urban environments, commercial application of driverless delivery robots will be feasible. But changing the operating environment is a pre-condition.

At the company level, the same approach is needed. Internally, there are clearly easier opportunities for robotic solutions, as Walmart and Gant illustrate, because companies can control many of the environments involved in manufacturing and warehousing. The UK-based robotic automation firm, Ocado, is one of a growing number of firms helping retailers to configure warehouses with layouts tailored to efficient operations of robots that are centrally coordinated through a computer.

Yet there’s still a lot that robots can’t do. Many strategically important tasks are customer facing, and require the application of AI technologies such as machine learning in order to operate. AI algorithms require data to train them and a key challenge is obtaining this data without compromising the customer experience; it is still a long way before we will see customer-facing robots that will be able to learn from encounters with humans in unstructured environments that they do not control or manage. The example of Microsoft’s chatbot, which Twitter users trained to spout racist nonsense in less than a day, vividly illustrates the problems AI programmers must negotiate.

So what does this mean for robots’ role in a post-pandemic world? Well, for starters executives should look for ways to support humans performing strategic tasks with existing technologies before investing in new ones to replace them. As the example of Stanford Medicine shows, simply introducing iPads in patients’ rooms allowed physicians to deliver remotely many aspects of care effectively during the COVID-19 outbreak without adding to the risk of infection. As a solution to the challenge of treating contagious patients, it was considerably cheaper than investing in robotics would have been.

Robotics technology has evolved significantly over the last few decades, and the current generation of robotic solutions do confer significant benefits. However, these solutions are geared to automate a narrow range of tasks in largely structured environments. So although the current crisis provides a clear opportunity to fast-forward robotic automation, businesses should be wary of pouring valuable capital into automating processes that are not strategic or are too difficult for the current generation of robots to perform effectively.