Augmentation Realized

Government leaders increasingly view global leadership in Artificial Intelligence (AI) as a national imperative. According to the White House’s Office of Science & Technology Policy, machine intelligence “holds the promise of improving worker safety, increasing productivity, and creating new industry sectors not yet imagined.” Given the stakes, the federal government is also positioning itself as a testbed and incubator for developing and maturing this technology.   

There's trepidation in the workforce, however; some federal employees foresee a wave of automation that ultimately replaces their human efforts. For example, 73 percent of federal employees say that their agencies have done a poor job communicating how technologies like AI will impact their role.

Research conducted by Accenture, The Coming Federal AI Productivity Boom, paints a more nuanced picture. According to the report, the federal workforce may experience over $500 billion in annual productivity gains by 2028. Instead of straightforward automation, however, machine-led augmentation that allows workers to achieve better performance will deliver the bulk of these benefits. This economic modeling projects that nearly half (49%) of federal workers' time could be eventually augmented, which is 5% higher than the average of the 15 industries analyzed.

“AI can be an intelligent assistant, adding resource capacity, reducing the workload, and driving new levels of citizen service,” said Britaini Carroll, principal director for workforce transformation, Accenture Federal Services.

Given the pace of AI innovation, agencies should begin now to ready themselves for this augmented future. However, the Accenture Federal Technology Vision 2020 found that while 85 percent of federal leaders believe that human-machine collaboration “will be critical to innovation in the future,” only 18 percent were preparing their workforce to meet that challenge.

As the government begins to explore the potential for AI to augment human workers, it's worthwhile to take a more in-depth look at what augmentation means in practical terms.

AI at work

In their book Human + Machine, Accenture’s Paul Daugherty and H. James Wilson used research from 1,500 companies to determine that the best outcomes are often achieved when workers and AI collaborate, using their distinct strengths to problem solve in new ways. For humans, these strengths include creativity, complex problem solving, empathy, and sensory perception. AI, meanwhile, excels at brute force computation and rapid pattern recognition.

In many cases, merely adding AI to an existing process delivers disappointing results, as the workflow often needs to be redesigned to take full advantage of this collaborative relationship. For example, using AI to preprocess applications to verify accuracy so workers can focus on subsequent tasks like adjudication requiring more subjective analysis. To better envision the future "intelligent assistant," it's helpful to look at examples of how AI already supports workers across various tasks commonly seen in the federal government.

Federal call centers use chatbots and other AI tools to handle or support a wide range of queries. At the most basic level, they may independently respond to simple questions. They can also be deployed to prepopulate forms using natural language processing so that the customer service representative (CSR) can focus greater attention on the caller. And in some cases, AI can be used to provide real-time coaching for the CSR or use machine learning to route calls to those best qualified to address specific issues.

“By offloading low-value work, people are empowered to be more productive and take on more complex, high value human tasks,” Carroll said. “They’re better able to keep up with demand, which is especially important at a time when demands on the federal workforce are growing substantially.”

Going a step further, Accenture researchers applied human and machine intelligence to improve medical coding speed and accuracy. It's a repetitive task, but one that is critical to supporting both billing and patient care – precisely the kind of high-volume, low-complexity job that AI does well. Historically, data scientists were responsible for updating the underlying knowledge-base. In this experiment, researchers created an interface that allowed registered nurses to directly interact with the AI, importing more complex concepts that further improved the system's performance. Simultaneously, the nurses focused more time and attention on the more complicated cases that required their domain expertise, decision-making skills, and critical analysis.

The data suggests that such human-AI pairings could have a substantial impact across the government. Accenture research shows that, depending on the worker's specific role, anywhere from 36 to 54 percent of a given federal employee’s tasks could be augmented by an AI partner.

"Government needs AI and automation to deliver the intuitive services that citizens have come to expect in the consumer marketplace," said Mary Anne Jeffers, analytics and modeling senior manager, Accenture Federal Services. “The national security community faces its own imperative to take advantage of AI, given the speed of its operations and massive amounts of data available today.”

Becoming more human

Business executives recognize that AI will drive workforce changes. According to Accenture’s 2019 Technology Vision report, 43% of executives report more than 60 percent of their workforce will move into new roles within the three years, requiring substantial reskilling due to technology's impact.

What’s clear is that the advent of AI will push federal agencies to rethink how people do their jobs and how they will interact with their emerging digital counterparts. The challenge? Currently, just 22% of federal executives report that they have inclusive or human-centric design principles in place to support human-machine collaboration.

On the federal side, training is needed to help people achieve a higher proficiency level in their current jobs. As AI takes on more of the mechanical tasks associated with government work – from sanitizing workspaces to confirming data on forms – the workplace attributes that are uniquely human will come to have even greater value.

Complex reasoning, creativity, social and emotional intelligence: All these will rise to the forefront in an AI-augmented workplace. “As people focus more on human-centered tasks, they will need to be able to relate to others. That requires increased reskilling in interpersonal communications, emotional intelligence, and trust-building,” Carroll said.

Take, for instance, the worker who puts fact-checking tasks in the hands of AI so they can dig deeper into a benefits assessment. “That person needs to able to communicate at a high level,” Jeffers said. “They need to have empathy, on the one hand, and they also need to ensure that this is not some fraudulent application. That kind of cognitive reasoning is going to be very important in moving forward.”

In this emerging environment, humans will also have a unique new role to play as the trainers of their machine partners. Subject matter experts will need to leverage all their experience, intuition, and insight to teach the AI. To some degree, they will also be the ethical gatekeepers, the ones who review the outputs of AI to ensure equity, fairness, and accuracy.

“From an ethical perspective, AI will be everyone's responsibility,” Jeffers said. “Our society is shifting, becoming more conscious of other people's experiences and about our own biases. As we develop these AI products, the expertise of federal workers will be needed to oversee those processes from multiple different perspectives.”

Going forward

Agencies’ leaders can take several steps now to prepare the workforce for a shift toward augmentation. They can lay the groundwork today by regularly highlighting examples of AI empowering workers and underscoring the enduring value of uniquely human skills.

“This starts with leadership working to create a culture that is supportive of these changes. People need to hear that AI isn’t coming in to replace them, that it is coming to support them, to help them do their jobs better,” Jeffers said.

Agencies can begin to build the teams that will drive AI implementation: The computer scientists, data experts, and subject matter experts who together will make AI a reality. Moreover, they can begin developing an enterprise-wide framework for AI, an organizational-level strategy for future deployments.

"In our research, we've seen people do 'one and done’ – one test, one use case,” Carroll said. “It’s more effective in the long term to have a more comprehensive AI strategy, with a leader who can bring together all the relevant workforce stakeholders for prototyping, iterative experimentation, and eventual scaled deployments.”

A comprehensive approach can help in these early days to establish the ground rules of AI usage and to ensure various AI-driven efforts are in sync across the organization. All this, in turn, will help to address the crucial human element by assuring workers of their central role in the organization and by laying out in clear terms the benefits of augmentation.

“The message to the workforce is: AI is not going to take your job. AI will start assisting with the parts of your job that are repeatable and standardized,” Carroll said. “That will allow you to work on the activities that rely on your unique human judgment and expertise."

That framing elevates the worker in the equation. It emphasizes that human knowledge and skills will be more valuable than ever as federal workers begin to partner with machine intelligence.