Over the last year or so, we have begun to see a discernible shift in the public discourse about how automation will affect work. From the early, often dire, predictions about how many jobs will be destroyed by automation, we are now seeing broader recognition that automation affects the component tasks of jobs in more nuanced ways. As my co-author John Boudreau and I wrote about in our book Reinventing Jobs: A 4-Step Approach for Applying Automation to Work (HBR Press, 2018), automation affects tasks in one of three ways; substituting, augmenting or creating human work.
To strategically understand and optimize work automation organizations and their leaders must: (1) Deconstruct “jobs” and reconfigure the work; (2) Fully understand the range of work automation options (including robotic process automation, cognitive computing and social robotics); (3) Adopt a systematic step-by-step approach to identifying and executing on work automation opportunities.
Start with the Work, Not the “Job” or the Technology
Applying automation requires thinking outside the boundaries of traditional jobs, units, hierarchies and processes. The “organization” must be reconsidered as a hub and capital source for an ecosystem of work providers. Those “providers” include AI and automation, but also include “human” sources such as employees, contractors, freelancers, volunteers and partners. As noted above, the optimal combination of these providers seldom appears if you frame the question as, “in which jobs will AI replace humans?” Perhaps the wholesale replacement of humans by AI or robots is 50 years away for some “jobs,” but when you look within those jobs the actual effects are substantial and will occur much faster. Optimal decisions about AI and automation reveal themselves only when you deconstruct and reconfigure the work elements within the jobs.
There are many approaches to deconstructing and categorizing the component activities of jobs, but we believe they can be illustrated with three fundamental work characteristics:
Repetitive vs. Variable Work
Repetitive work is often predictable, routine and determined by predefined criteria while more variable work is unpredictable, changing and requiring adaptive criteria and decision rules.
Independent vs. Interactive Work
Independent work requires little or no collaboration or communication with others, while work performed interactively involves more collaboration and communication with others and relies on communication skills and empathy.
Physical vs. Mental Work
Physical work is work that is primarily manual in nature, requiring manual dexterity and, often, strength while mental work requires one’s cognitive abilities.
Apply the Relevant Form of Automation
The three forms for work automation to consider are robotic process automation (RPA), cognitive automation, and social robotics.
RPA automates high volume, low complexity, routine administrative “white collar” tasks — the logical successor to outsourcing many administrative processes, further reducing costs and increasing accuracy. Optimizing RPA can only be done when the work is deconstructed. For example, RPA will seldom replace the entire “job” of a call center representative. Certain tasks, such as talking a client through their frustration with a faulty product or mishandled order will, for now, remain a human task. Other tasks such as requesting customer identification information and tracking the status of delivery are optimally done with RPA.
Cognitive automation takes on more complex tasks by applying things like pattern recognition or language understanding to various tasks. For example, the Amazon Go retail stores around the country have no cashiers or checkout lanes. Customers pick up their items and go, as sensors and algorithms automatically charge their Amazon account. Automation has replaced the work elements of scanning purchases and processing payment. Yet other elements of the “job” of a store associate are still done by humans, including advising in-store customers about product features.
Social robotics involves robots moving autonomously and interacting or collaborating with humans through the combination of sensors, AI, and mechanical robots. A good example is “driverless” vehicles, where robotics and algorithms interact with other human drivers to navigate through traffic. Deconstructing the “job” reveals that the human still plays an important role. While the human “co-pilot” no longer does the work of routine navigation and piloting, they still do things like observing the driverless operation and stepping in to assist with unusual or dangerous situations. Indeed, it is often overlooked that the human co-pilot is actually “training” the AI-driven social robotics because every time the human makes a correction, the situation and the results are “learned” by the AI system.
Adopt a Systematic Approach
Achieving the optimal combination of humans and machines is easily accomplished with the right approach, framework, and tools. The following checklist will help leaders ensure success in their quest
- Identify the opportunity
- Is there an opportunity to significantly reduce cost through automation?
- Are there new or emerging capabilities that you can develop through automation?
- Are you struggling to find talent for critical work areas where deconstruction and automation of certain tasks may make it easier to get the remaining aspects of the work done (either by existing staff, contingent talent, or more easily accessible talent)?
- Can you select a pilot location to experiment with automation?
- Understand the work
- What are the characteristics of the work and the relationship between the performance of a particular task and the value derived (this is defined as the Return on Improved Performance (ROIP))?
- If grouped into a job, can these activities be separated with minimal breakage?
- Is there valuable connective tissue between activities that will be critical to maintaining? Are there means for doing so that go beyond aggregating them in a job?
- What is the expected return from automation (productivity, speed to capability, cost, risk, etc.)
- Apply automation
- What insight does your analysis of the role and type of automation yield?
- What is the relevant type of automation (robotic process automation (RPA), AI/cognitive automation or social robotics?
- What is the role of automation (to substitute, augment or create new human work)?
- How will you acquire this automation? Is it available as software as a service (SaaS), as is the case for numerous RPA and cognitive automation solutions? Or, in the case of social robotics, can you lease the equipment from a manufacturer?
- What customization of the automation is required? Who will do this and how?
- Ensure proper oversight and governance
- Which functions of the organization (HR, IT, procurement) will need to get involved as you apply automation?
- What will be the role of each stakeholder? Who will have primary responsibility for coordinating work in the newly automated workplace?
- How will you train managers and employees to work with your new automation solutions? What new or different skills are required?
- What are the security implications of automation (think potential new cybersecurity vulnerabilities)?
- Measure your ROI
- How did the actual return from automation compare to the expected return?
- What factors caused the deviation?
- What do you need to change as you move from pilot to full-blown implementation?
Source: Forbes – Leadership