The Data Center Journal | Automation Spawns New Jobs and Takes Old Ones

This article was written by Greg Council and first published on August 16, 2017 in The Data Center Journal and an excerpt is provided below:

Automation—whether from robotic process automation (RPA), traditional expert-based systems or new deep machine learning—is causing both excitement and anxiety regarding how it affects economies and jobs.

What’s the real driver of job loss? Is it globalization? Automation? Technological innovation? It’s all three and more when assessing the actual reasons for changes in employment. Although you can’t go far on the Internet without finding a doom-and-gloom article blaming automation for future mass unemployment and crying for technology and robot taxes, the reality is much more nuanced.

Automation isn’t all the same. There are different types of automation with diverse impacts. Some types will completely replace humans and others will make work more efficient (and maybe more enjoyable) for us.

For organizations of every size and in all industries, automation will affect operations both in the near future and beyond. So, how best to plan for it and exploit the opportunity rather than suffer competitive disadvantages by falling behind?

Automation and Job Loss

Automation is less about job loss than about consistency, error reduction and access to completely new capabilities. With these advancements, automation is a critical factor in opportunities for cost reduction, risk reduction and greater overall output. An even more interesting point is that countries with the highest levels of automation don’t necessarily have a corresponding increase in job loss. A common illustration is Germany, which has three times as many manufacturing robots as the United States, but manufacturing jobs declined by less than half of what we encountered in the U.S.

Although automation is essential to continued economic growth and competitiveness, it doesn’t necessarily translate into job loss. That’s good news to any organization that’s looking to improve through automation. But what’s the impact on occupations outside of the manufacturing sector where we’ve already witnessed disruption?

Types of Automation: Big Force Behind Labor

Today, the U.S. is the second-largest manufacturer behind China, but its GDP share is dwarfed by the services sector, which has been unaffected by robots and other automation until recently. A meta study conducted by PwC concluded that some countries’ service-oriented industries will be affected more than others, primarily owing to the nature of tasks and the percentage that are repetitive or rule based.

As McKinsey & Company points out in a recent article regarding automation, it’s more important to look into specific details of individual activities than to examine overall occupations or industries when evaluating impact. In its research, published in January, the company estimates that 5 percent of occupations could be completely displaced by automation, while almost every occupation could be affected by some level of automation. It’s insufficient to examine only whether or job will be subject to automation—it’s more important to look at the level of automation that results.

When deconstructing types of tasks and their potential for automation, we can generally consider four categories:

  1. Simple-task automation. This rote task consists of a few steps where the number of potential variances is low (e.g., a few deterministic rules). Sorting different known documents by type is one example of a simple task. It’s simple because the documents are all known and the types are finite.
  2. Complex-task automation. Complex tasks consist of multiple steps, each one potentially consisting of more options. The key is that all of the options are known. Claims adjudication is an example of a complex task. In claims adjudication, many different policies can determine whether a claim is allowed, rejected or adjusted. So the task is more complex, but the different policies are all known and definable. Once staff members are trained, they can operate with minimal disruption and little uncertainty.
  3. Analytical-task automation. These tasks require review and synthesis of a range of data to produce a particular answer. As such, the data variance can be higher and the calculations much more complex. An example is calculation of a company’s loan-default risk. Different loans carry different terms, and companies have different risk factors based on financial status and industry—to name a few.
  4. Cognitive-task automation. This set of tasks is generally less defined or undefined. The ability to automate on the basis of predefined rules or even with more-powerful and general machine-learning algorithms is extremely difficult or impractical. An example is customer interactions for services that include a lot of information and variation. For instance, the interaction between a car owner and a mechanic is a largely undefinable process that can include many different types of communication and analysis depending on the problem.

Most occupations mix the above tasks at different percentages of the overall workload. A data-entry operator might have 90 percent simple tasks, with the remainder consisting of complex tasks; an HR manager might have 90 percent cognitive tasks.

Several consultancies have created indices of occupations that would be the most affected on the basis of a similar breakdown of task complexity. The ability to break down the types of tasks in a given role will help an organization better understand how and where the impacts are likely to occur and how to exploit automation for competitive advantage.

Impeding and Accelerating Factors

Although we’re all enamored by AI-based services such as Alexa and Watson, automation in an enterprise is more complicated than plugging it in and asking for the weather report. It doesn’t exist in a vacuum, free to take over tasks wherever applied. Adoption involves friction. Generally, we can consider three major factors, although many more exist:

  • Complexity of technology. For even simple, repetitive tasks such as data entry, we continue to see companies hold onto data-entry staff rather than fully use automation. For instance, a large service provider may choose to continue to employ a large staff to perform data entry because managing a human workforce is straightforward and a known entity, whereas adopting automation technology isn’t and comes with risk. Even organizations that have adopted automation may not completely avail themselves of efficiencies to avoid complexity. For instance, that same service provider, using automation, typically elects to carry staff to verify the results of automation because measuring and ensuring accuracy is seldom easy and requires lots of effort from higher-wage staff. Simply put, if the technology in question is more complex than the human alternative and carries additional risks, organizations will be slower to adopt it.
  • Supply and demand. Companies evaluate investments on the basis of expected return. In the services sector, adoption of automation was staved off by the ability to move labor supporting those services to lower-cost regions, aided by improvements in telecommunications and computing technologies. Doing so allowed for near-real-time responses regardless of time zone. Businesses are unlikely to invest in complex, expensive automation if the equivalent market rate for a data-entry operator is less expensive overall.
  • Impact on workflows. We’ve established that in most cases, automation will affect only a portion of most roles. The actual impact will occur as a complement to existing occupations, not a complete replacement for them. Therefore, the automation must seamlessly fit into human-oriented processes with minimal disruption, or the organization risks a workforce revolt. For example, if companies implement automation for compiling data to comply with the Sarbanes Oxley Act, CFOs and their reports will be involved. If the process requires a major effort to condition data, import it into the system and provide output, the team may elect to continue to manually prepare reports, judging that the effort to prepare data is more costly.

Final Impact: Proportioned Automation

Although automation will eliminate some occupations entirely, the largest impact will be on portions of roles. Benefits will come from expanding the ability of current staff to focus on high-value work, potentially yielding growth in some occupations. If staff can focus on high-value tasks, demand will follow. “If I only had more hours in the day” becomes a thing of the past. An example provided by McKinsey is growth of cashier jobs in grocery stores over the same time period as adoption of more sophisticated point-of-sale systems that included bar-code scanning. Similar growth took place among bank-branch staff even while ATM adoption took place.

Even for occupations consisting of simple and complex tasks, automation has benefits. These types of jobs typically represent lower-value activities that are often offshored, but they’ll potentially reappear in the U.S. as hybrid human-machine processes—enabled by automation. The number of jobs involving these types of tasks will never be the same as the number before automation, though.

Improved Productivity

Additionally, automation will improve productivity. Since many occupations will only have partial ability to automate, the remaining time can be devoted to performing more of the complex work that cannot be automated. In health care, many activities incur large costs. For example, the current state of computer-aided detection technology is almost on par with that of an experienced radiologist when it comes to identifying suspicious areas in a medical image. Yet examining and interpreting an image is only a small part of the value of this profession. These professionals also are examining a host of other information to derive an overall analysis of a patient’s health, and they provide therapeutic services. They consult with other medical providers and with the patient. What would the benefit be of allowing a radiologist to focus on more value-added services or to support a larger patient load?

Automation is now the big force behind labor. The positive result is that it ultimately will mean more jobs here in the U.S. that traditionally would be lost forever, as well as an expansion of productivity by allowing a larger allocation of workforce time to focus on high-value activities.

About the Author

Greg Council is the VP of Marketing and Product Management at Parascript. He specializes in bringing products to market in the document-capture, enterprise content-management and business process-management markets.