Advanced Computing in the Age of AI | Friday, September 17, 2021

Study: AI’s Not Just in the Future of Work – it’s Big Now 

Pundits trumpet AI as the future for U.S. employment, but new research indicates AI's prevalence on the job is already here. How do we know? Because workers told us.

Data based on a survey of U.S. workers in June indicates more than 32 percent are already exposed to some form of AI in their jobs. And not surprisingly this prevalence is expected to grow; of the workers who are not exposed to AI at work today, 6 percent say their departments have plans to roll out at least one form of AI tool by mid-2019.

AI's influence in the workplace is appearing in subtle ways, not extreme ones, and this influence is best detected in data collected directly from workers on the ground, rather than from macroeconomic forecasts. We surveyed more than 10,000 U.S. workers to understand the time they spend on specific tasks, the AI technologies they work with, and the technologies they plan to deploy next year to help with those tasks. The survey sampled workers from 19 of the 20 Census Bureau NAICS codes and all of the Bureau of Labor Statistics' top-level occupational codes.

Understanding time spent by task also allows us to calculate the dollar value of the efficiencies AI creates. (In addition to collecting time-on-task data, we also collected salary and compensation data per respondent.) We estimate the dollar value of efficiencies AI could provide the U.S. economy in 2018 to approach $900 billion – the equivalent of 15 percent of total U.S. payroll.

That’s a big number, but AI technology vendors will see only a fraction of that this year. One reason for this is that – obviously – some of the value vendors create must be passed on to their clients, or else implementing the technology is not worth doing. But the survey indicates that proving AI ROI also plays a role. Two industries with the greatest need for AI – the process-intensive health care and government services sectors – are among the slowest adopters. These industries burn $56 billion per year in manhours on administrative tasks alone; plus, time spent on other tasks that AI can make more efficient, like data analysis and data entry, accounts for another $30 billion. But these industries trail early adopters – like the information and manufacturing industries or construction – which are poised to make aggressive investments in 2019.

If AI in the workplace is so prevalent, why does it still feel like futuristic science fiction? Largely because people tend to think of AI's role in the workforce in extremes. The bulk of news reports and white papers forecasting AI's impact on the economy typically focus on the elimination of certain livelihoods; a long over-played scenario is the demise of taxi- and truck-driving subsector at the hands of autonomous vehicles.

In reality, today's low-hanging fruit for AI vendors involves easing the burden of administrative tasks – time tracking, meeting scheduling, travel planning and similar non-core functions – not wiping out entire professions. These tasks take up a lot of employee time and they are common across nearly every industry. However, the survey results imply that AI tools with more strategic applications – like scenario planning – will also gain steam in 2019.

In part, AI adoption mimics the consumerization of tech we witnessed in the early 1980s. Enterprise technology makers were quick to create breakthrough after breakthrough in computing technology, but until the consumer-facing applications made it "real" to the common user, adoption was slow to take off. We expect a similar scenario to play out today with AI. Chipmakers like Nvidia and cloud computing providers like AWS are making AI development possible. But until more companies like (in ad buying) or Shutterfly (in HR and talent strategy) take the building blocks and offer technology the average business user can understand, it will be hard for many budget-holders to see the value.

An AI “x-factor” to watch: natural language processing and voice recognition-based solutions. Although our survey indicates moderate corporate adoption of these technologies compared with document classification and document creation solutions, Amazon, Google and Facebooks’ aggressive push of voice-AI into consumer markets this holiday season will undoubtedly open doors for business applications.

Data from the survey do not forecast an immediate negative impact on employment due to AI. Although the efficiency equivalent of 15 percent of total U.S. payroll is a scary number, we found that more than 13 percent of workers surveyed said they spend so much time on tasks that could be automated that they are missing key business goals and deliverables. Although AI's long-term impact on employment is much more difficult to predict, in today's business climate – with many overworked professionals in high-value industries like health care and professional services, and unemployment rates at historic lows – AI is more likely to help the majority of workers than hurt them.

To see the 2018 AI Preparedness Survey, please go to: Optimized Workforce.

Craig Desens sits on the Board of Advisers for Optimized Workforce. He has held leadership positions at analytics companies, including CIO and COO roles at IPG's NSA Media subsidiary and SVP, customer support at Networked Insights.

Add a Comment