Step One: Deconstruct Your Job

Deconstruct your job into its tasks and consider how each will evolve. To find your work tasks, start with your job description and its tasks, outcomes, and competencies. Also, consider how you actually do the work, particularly if that’s different from the formal job description. Describe the additional tasks and add them to your list.

Evaluate each work task against the three characteristics of work elements, to see which may evolve away from traditional employment or toward automation. Are the tasks:

• Repetitive or variable?
• Independent or interactive?
• Physical or mental?

The more repetitive, independent, and physical the task, the more likely it can already be automated or will be soon. Even mental work, if it’s independent and repetitive, can be automated by RPA and AI. Physical work will probably be automated by robotics.

Now, write the job description as it might appear in two, five, and ten years. Remove the tasks where automation will substitute for humans. Keep those tasks where automation will augment the work and consider how your work will change with augmentation. Finally, what tasks are humans likely to do for a long time? Which of those tasks are regular employees likely to do, and which might be done through arrangements other than employment?

Once you deconstruct and reinvent your job, you will probably find that your reinvented future job contains a fraction of the tasks that you do today. What work elements might be added to your job as it evolves (i.e., new work that will be created)? Can you reconstruct a job from the work that remains plus the new work? How does that align with your unique skills? Could you do some work through freelance platforms, gigs, contracts, or other engagement models either in your organization or elsewhere?

The Economist describes how future work will combine human workers, automation, and freelance work platforms:

Step Two: Assess ROIP

For each current and reinvented task, how does improved performance create value? The four ROIP curves we described in chapter 2 can help you map the payoff of the work. Figure 7-1 shows the four curves; the vertical axis is the value of performance, and the horizontal axis is the level of performance.

You can often find clues to the ROIP of work tasks in discussions with your organization about your performance, key performance indicators, and so on. Conversations about performance, goals, rewards, and development often rely on a shared understanding about the tasks that make up the job, and how those different tasks pay off.

With ROIP identified, consider the implications for work and automation. For the tasks that add ROIP by avoiding mistakes, could robots, RPA, or cognitive automation better reduce human error? If the tasks with error risks are automated, might you be free to spend more time on the tasks that produce incremental value?

For the tasks where different performance approaches add very little value beyond a certain minimum standard, is it possible that AI could learn the best way to do the work? Or will AI help train or nudge workers toward a common way to perform? If you could expect this work to be done consistently, would that allow you to stop wasting time on unproductive variations?

For tasks that add incremental value, will automation help you move up and to the right on the ROIP curve? For example, if you add incremental value when you interact with customers, will automation improve your interactions by giving you better information about the customer? Will AI essentially make you and other workers equivalent to today’s top performers? Or, is your greatest value in training the AI on the most successful customer-interaction patterns?

Finally, for tasks that create exponential ROIP, how will automation create whole new kinds of work? For example, if your value is in understanding the subtleties of how equipment operates or how parts of a process integrate, then when automation takes on the tasks of operating the equipment or monitoring the process, will your new work occur in a cockpit, where you operate and diagnose whole arrays of machines and processes, just like the oil drillers we discussed earlier?

Step Three: Identify Automation Options

Now, consider the three types of automation (in chapter 3): RPA, cognitive automation, and social robotics. Determining which types of automation will apply to your work will help you understand the likely speed and consequences of automation for you. Remember that in most real situations, several types of automation will converge.

RPA applies best if your job has mental tasks with one or more of the three Rs: repetition, redundancy, and risk. RPA needs no programmed intelligence, so it is typically the cheapest, quickest, and easiest to apply. If your tasks are amenable to RPA, you can probably assume they will not be done by you or other humans.

Human workers and their union or other representatives often resist RPA, fearing that they will be displaced. That is a very real possibility. However, the long-term solution will seldom prevent the new technology. Rather, the optimum and most socially beneficial answer is likely to emerge as workers and leaders openly and regularly discuss the implications, and reinvent the jobs before humans are displaced. A recent Chicago Tribune article described a Wisconsin factory where robots don’t replace humans, but they fill in the gaps caused by worker shortages.⁷ On an assembly line intended for twelve workers, “two were no-shows. One had just been jailed for drug possession and violating probation. Three other spots were empty because the company hadn’t found anybody to do the work. That left six people on the line jumping from spot to spot, snapping parts into place and building metal containers by hand, too busy to look up.” Despite increases in wages and efforts to recruit qualified and stable workers, the shortage persisted.

RPA and social robots provided a win-win solution: “In earlier decades, companies would have responded to such a shortage by either giving up on expansion hopes or boosting wages until they filled their positions. But now, they had another option. Robots had become more affordable. No longer did machines require six-figure investments; they could be purchased for $30,000, or even leased at an hourly rate.” Supervisors explained that the robots would not take anyone’s job, but make up for jobs they couldn’t fill. One worker, reliably doing the job for years, tired of the constant turnover and empty work spaces that required her to work harder. When one employee said, “They are taking somebody’s job,” she replied “No, they are not . . . It’s not a good job for a person to have anyway.”⁸ Cognitive automation can do things like recognize patterns, understand language, and learn rules. Which of your tasks does these things? If the patterns are simple, the languages are well known, and the rules are fixed, then cognitive automation or AI may substitute for humans, or already has. The move to automation is even faster when cognitive automation can be combined with sensors, such as the cameras equipped with AI that know they are photographing.⁹

On the other hand, the more complex or obscure the patterns, languages, and rules of your task, the longer cognitive automation will take to master them. Indeed, in the interim between fully human and fully automated, the new human work is training the cognitive automation. Human riders train the AI that supports self-driving cars. Every time the human corrects the AI, the program learns. Perhaps your work will evolve from doing the task to training the AI to learn the task. Uber built a fake city in Pittsburgh, called Almono, populated entirely by self-driving cars. In the fake city, automation designers can simulate things that they can’t test in the real world, such as mannequins darting out front of vehicles. In addition to vehicle testing, an important goal is to train the human drivers who will eventually accompany the vehicles on actual roads: “The program that trains vehicle operators is rigorous. It takes three weeks to complete and requires trainees to pass multiple written assessments and road tests.”¹⁰

Of course, cognitive automation may soon teach itself. In October 2017, the AlphaGo Zero AI system achieved what some have called a “singularity” in which it learned to play the board game Go with no human interaction. Designers simply programmed the game with the rules and objectives, and then instructed it to play by itself, over and over, and very quickly. An earlier AlphaGo system had required programming with the past games human experts had played. That first system defeated a world champion player in March 2016. The new system, playing only by itself, beat the original AlphaGo.¹¹ So, be alert to continuing evolution. Perhaps, in the near term, you will teach automation, like the original AlphaGo, but eventually automation like AlphaGo Zero will learn on its own, so you will need to reinvent again.

Social robotics involves robots that move and physically interact with humans. The robot Baxter is a cobot that literally works next to humans to do things like loading, machine tending, and materials handling. Fast Company described cobots that greet guests at the Henn na Hotel (translation “strange hotel”), and make coffee and clean rooms at the YOTEL New York; “Botlrs” at Starwood Hotels that use the elevator without assistance to deliver amenities to guests; an OSHbot at Lowe’s stores that locates items for customers, and an AI security guard called Bob.¹² At the MIT Sloan School of Management, staff who are working from home can attend meetings, as a robot.

A Financial Times article points out the need for workers to constantly analyze how their jobs will be reinvented.¹³ It describes a facility with a traditional human-worker assembly line on one side and a cobot-enhanced line on the other:

Such low-cost cobots now allow small local factories that could not compete before to operate competitively and retain jobs for humans. Yet, the article also noted that many companies refused the reporter’s request to see their cobots in action, perhaps fearing adverse publicity. Tony Burke, assistant general secretary of the Unite union, noted that “job losses could be horrendous in some areas . . . but the problem is that nobody really knows.”¹⁵ The solution is not to deny the effects or to hide from the press. Progress will require difficult, candid, and open conversations about the specific effects of automation at the work-task level. The framework in this book can help define and guide those conversations, to move beyond simplistic questions focused on job losses to a more nuanced conversation about the optimal work-human combinations, with society, workers, unions, companies, and communities defining “optimal.”

Step Four: Optimize Work

The first three steps, applied to your work and the work of your colleagues, prepares you to have the right conversation about how the work will evolve and how to prepare for it. The key is to distinguish and integrate the elements of our framework: work tasks, ROIP, automation type, and automation role

John Donahoe, the CEO of ServiceNow, put it well: “There’s this assumption that it’s going to be people or robots, all or nothing. My experience is that it doesn’t operate that way. It’s automating part of the job, but not the full job. Repetitive, manual work—no one who’s doing it is really enjoying it. Technology replaces and creates. It replaces manual work and creates new opportunities—new tasks, if you will. And productivity creates growth, which creates new kinds of work. It is a virtuous cycle. It’s so easy to talk about it in binary terms. I just don’t think that’s the reality.”¹⁶

Step Five: Navigate the Organization

Figuring out how your tasks will evolve and how to reinvent your job is not the end of the conversation. Your work will exist in a larger organization that is also being reinvented. That organization will not be confined by the traditional boundaries, with employees inside doing the work and others outside waiting to join. Rather, the future organization will be a hub of work arrangements like the ones we discussed in the previous chapter: free agents, talent platforms, volunteers, alliances, outsourcing, robotics, and AI.

You must navigate this organization by finding the optimum way that your reinvented job fits and connects with those who are your clients and collaborators. Sometimes you will be a traditional employee, sometimes a freelancer on a platform, and sometimes you’ll take a tour of duty with an alliance partner, and so on.

You will face questions like these:

• What part of your future work should be kept in a traditional job?
• What proportion of future work should you pursue as a side gig through a work platform?
• What tasks will be substituted versus augmented by automation?
• How should you get the experiences and skills that will support your evolution?
• In what tasks and jobs should you become an expert, available to train AI or humans?

All the while, you must also be alert to the perpetual work upgrades that are the new reality of your work.

As a leader, you must create an environment where you collaborate with workers and other constituents to answer a parallel set of questions for your organization:

• What future work should be kept in a traditional job?
• What future work should come through gigs on a freelance platform?
• How should you give your workers the experiences and skills that will support their evolution (such as tours of duty, projects, social learning, etc.)?
• For what areas of work should you transform those that do the work to become the trainers of AI?

All the while, as a leader, you will also observe the perpetual work upgrades that are the new reality of your work. But, as a leader, you will also consider not just your work but related work and how to create the organization of the future.

The good news is that if organizations, nations, governments, and workers are empowered to work together, you are more likely to have the tools and structures you need. The new work ecosystem with constantly upgraded and reinvented jobs has the potential to empower workers, create boundless opportunities for careers and learning, and solve thorny issues of skill gaps and regional inequality. The bad news is that if workers and organizations navigate this new world in secret, without trust and transparency, then the work ecosystem risks being exploitive, opaque, and dysfunctional. That trust and transparency requires processes like “reinventing jobs” that offer a framework and language to candidly and clearly communicate about the coming evolution. (See the sidebar “Principles Guiding the Development and Use of AI.”)

As a leader, you can be a role model, supporting the conversations with opportunities, a framework, and the necessary information and safety net that empowers you and your workers to collaborate to optimize work automation. You can help your workers follow these steps. You can create a safe culture in which workers can tell you when they see the possibilities for automation to replace their tasks and where augmenting tasks with automation could significantly improve productivity. If workers think that such revelations will result in their dismissal or exploitation, they won’t share. If that happens, then you miss the chance for the productivity enhancements, and your workers miss the chance to evolve in advance. As challenging as the conversations will be, it is likely much more painful to make these transitions when everyone hides until forced to act.