Paul Whitelam, VP Product Marketing, ClickSoftware, puts across the case that in the race towards AI adoption we shouldn’t forget to see the value and importance of human input in the service cycle...

Like many industries, field service has seen an increase in the adoption of artificial intelligence-driven automation.

The benefits are many: improved efficiency, schedule accuracy, workforce productivity, responsiveness, cost savings and higher profit margins, and, importantly, happier customers.

Naturally, the onset of automation causes some anxiety in workers whose tasks are being handed over to AI. As with previous industrial revolutions, we’re not likely to find ourselves in a low employment high-leisure utopia. While the nature of work might change, plenty will remain to be done. Getting the full benefits of AI and machine learning still requires some human participation and a good understanding of who (or what) is best for each job.

People provide context

Service management solutions powered by artificial intelligence and machine learning can rapidly process high volumes of data to use as a basis for automated decisions. But when it comes to learning, machines can be a lot like humans— garbage in, garbage out.

When Microsoft launched its Tay chatbot on Twitter in 2016, few would have guessed that in just a day it would become a bigoted bully. The problem, of course, was that Tay was learning to converse by interacting with Twitter users, some of whom seized the opportunity to educate it on humanity’s worst impulses. Even with less shocking or inflammatory outcomes, AI learns from what it is shown and told. It’s likely to replicate bad behaviour if that’s all it’s shown.

AI-based tools can also provide simulations and modelling for multiple scenarios and highlight the interaction of various policy and process changes. For example, if the objective is the fastest response time available for every job, more technicians might have to be available for dispatching, increasing labour costs and decreasing utilization.

People must still define the process and priorities for automation to ensure your system optimizes for the right business goals. While intelligent computing power can grease the wheels of daily service operations, the real value comes from informing businesses to foster improved decision making.

Managing the unique and unusual

While humans can grow bored with the rote and routine, machines have yet to complain. Tasks that are repetitive and predictable are best handled with automation.

AI can manage most routine and ordinary tasks – chatbots, scheduling, appointment confirmation, routing, showing a mobile worker’s location and travel path to a job, it can even reassign and redistribute jobs around disruptions, addressing unplanned work with urgency. One UK gas utility can dispatch engineers to address a leak emergency in 13 seconds from the initial customer call—without human intervention.

AI can use a variety of inputs to increase schedule and travel time accuracy and optimize in real time, but what happens when you just don’t have the data?

One of the challenges faced by self-driving car producers is how to navigate remote areas, especially with routes that lack landmarks or distinguishing features.

Too few inputs can stump the machine. There is also additional context in some situations that will not be gleaned from data analysis, and impact from factors that perhaps are not being measured.

Hands off, humans

Applying new technology to solving problems in old ways yields minimal benefits, if any. Field service organizations see the greatest benefits from automation after reviewing their processes, KPIs, and business goals to leverage exactly the kind of data processing and analysis they didn’t have before.

They guide machine learning by providing good and plentiful data, filtering out the unimportant, and prioritizing the right goals. Specificity is key.

AI will do exactly what you tell it to—including replicating inefficient processes or making dubious decisions to optimize for a single outcome.

For example, prioritizing the shortest possible wait times for a technician to arrive could result in overstaffing and idle time—costing a lot of money.

Use projections and simulations to see how various goals interact to find the optimal balance, and remember that instructing your system includes telling it what not to do.

The vision of zero-touch service scheduling and dispatching enabled by AI and the Internet of Things is increasingly becoming a reality for service providers. Resist the temptation to interfere when unnecessary so you can give the machine a chance to learn, and reap the full benefits of increased productivity and efficiency.

What can your employees do with the extra time in their day? Focus on the people stuff, of course: training and coaching, brand ambassadorship, cross- and upselling, remote support—you name it.

There is still plenty for humans to do in the increasingly automated field service world, and it’s the work that relies on person-to-person connections and trust. While your people are improving service quality and strengthening relationships with colleagues and customers, trust that automation can handle the rest.

Be social and share