Artificial intelligence will soon be applied to a significant share of the work that U.S. government workers perform. Our cross-industry model projects that nearly half (49%) of the time that federal workers spend on current tasks could potentially be augmented. This is 5% higher than the average for all 15 industries analyzed. (Note: the public sector is classified as a stand-alone industry in our model). Indeed, the only industries with a higher share of workers’ time exposed to augmentation are “education” (63% of workers’ time), “human health and social work” (52%), and “financial activities” (51%).
In 2021, in our baseline investment scenario, AI is expected to be used on tasks that consume about 10% of federal workers’ time; in 2028, we forecast that figure to jump to 30% (Figure 2). But AI won’t affect all federal jobs equally.
For a closer look at how AI will reshape the federal workforce, we modeled the percentage of workers’ time affected in different types of jobs (depicted as circles in Figure 3) if investment in AI approached the intensive scenario described previously. (We focused on underlying tasks—not on, say, job titles—because many roles involve a range of tasks). Figure 3 shows, for example, that workers who are employed in “technical equipment maintenance” jobs may have up to 36% of their time affected by augmentation, that workers in “process and analysis” jobs may have up to 44% of their time augmented, and that workers in “science and engineering” jobs may have up to 54% of their time augmented.
As Figure 3 makes clear, AI will soon affect the work of many federal employees, albeit unevenly. But are workers prepared for the big changes ahead?
Accenture’s previous interviews with federal executives offer some cause for concern. A 2019 survey, for instance, found that 35% of federal executives believe that up to half of the entire federal workforce could be made redundant by AI if their skills are not upgraded. In the same survey, more than half of federal agencies that already use AI reported that they lack “strong expertise” in such technologies.
The forthcoming Accenture Federal Technology Vison 2020 survey found that 85% of federal executives believe that “collaboration between humans and machines will be critical to innovation in the future.” However, the same survey found that only 18% were preparing their workforce to interact with collaborative, interactive, and explainable AI-based systems.
These findings, among others, indicate that more must be done to make the most of the government’s surging investment in AI. Indeed, our model suggests that, under the current baseline investment scenario, the U.S. government could see up to $66.5 billion in AI productivity gains vanish in 2028 if a “human + machine” workforce is not in place—i.e., one where workers are retrained to complement machines, allowing both to work together in new ways.