AI in the Workplace: Predicting the Unpredictable
Part of a series | Insights in Action Series
The Insights in Action virtual summit, hosted by ADP, explores the various ways data and artificial intelligence (AI) are impacting businesses today. For the event's opening session, Stanford professor Eric Brynjolfsson explores how AI is heralding a second machine age and what it may mean for the workplace.
The future of artificial intelligence (AI) is full of unpredictability. The technology centers on developing computer systems that handle problem-solving tasks in a way that mirrors human cognition, and its capabilities are reaching further every day. And while some people picture AI in the workplace as a handy tool that will help humans work smarter and faster, others are certain it will replace them.
As evolving AI meets established human talent, business leaders are feeling the impact — yet it's difficult to accurately predict the scope and scale of this impact on the workforce.
ADP recently hosted an Insights in Action virtual summit that brings this question and others surrounding data and AI for businesses to the forefront. In his keynote session, "Predicting an Unpredictable Workplace," Eric Brynjolfsson, a professor at the Stanford Institute for Human-Centered AI, discusses how AI is undoubtedly heralding a second machine age and what AI integration may mean for organizations.
Technology that bends the curve of history
Brynjolfsson says that one of his favorite questions to ask at dinner parties is what events each person thinks shaped the course of human history. "You'll hear about battles, about kings, about plagues," he says. "But for me, it all comes down to technology."
He points to the average market value of goods produced — the gross domestic product (GDP) — over time to illustrate what he means. For millennia before the Industrial Revolution, GDP rose slowly, but it wasn't enough for the general population to do much more than survive.
But with the creation of the steam engine, GDP skyrocketed. It was technology that led to widespread changes in how humans operate because steam engines made it possible to transport anything — people, animals and goods — anywhere train tracks were built. It wasn't the first invention to help streamline industrial processes, but it changed the game because it could streamline any industrial process. According to Brynjolfsson, that quality made steam engines the world's first general-purpose technology (GPT).
How AI is poised to be an influential GPT
Brynjolfsson posits that AI meets these three criteria necessary to qualify as a culture-changing GPT:
It's pervasive
Artificial intelligence offers capabilities including classification, labeling, perception, prediction and diagnosis. As Byrnjolfsson, Rock, and Syverson note in a 2017 work, these functions are "core to a broad range of tasks, occupations and industries."
It's improving
The fuel that powers AI tools are machine-learning algorithms, which are a sequence of problem-solving steps. These algorithms let AI improve over time — as more data is ingested and analyzed, outputs become more accurate and reliable.
It's innovative
General-purpose technologies must also drive innovation. In the case of AI, perception and cognition are two areas of ongoing innovation. Perception includes applications such as voice and vision recognition, while cognition speaks to advanced problem-solving.
Mind over muscles: The second machine age
The first machine age replaced human muscles with machinery. From steam engines to assembly lines to electric power, machinery made it possible to improve productivity while reducing the amount of human effort required. For Brynjolfsson, AI represents the start of a second machine age, where tools are used to augment the human mind.
This shift has become necessary in recent years thanks to the sheer volume of data generated by digital tools and technologies. Even highly-trained human staff can't effectively navigate the velocity and variety of this data, and the nature of human intelligence means employees simply aren't suited to detail-oriented, repetitive jobs that require no errors.
AI, meanwhile, is ideally suited to these tasks. Computer models can be trained to collect, curate and correlate data — and the more data they ingest the more accurate they become. This makes it possible for humans to handle the big-picture thinking while AI takes care of the details.
In practice, AI is already being applied by businesses in the form of chatbots that enable customer self-service and natural language processing tools capable of understanding employee queries in plain language and responding in kind.
Where AI fits in the new workplace
For organizations, it's not enough to know that AI works. They want to know about how AI will benefit the workplace.
But this is where things get unpredictable. On one hand, AI offers the promise of improved data analysis in a fraction of the time required by human staff. On the other, it creates worry about replacing human staff with AI.
Brynjolfsson notes that while AI is improving daily, it won't be a human substitute anytime soon. In support, he points out how some experts predicted that AI would rapidly outpace human radiologists because of its precise, repetitive nature. But it turns out that struggle with certain aspects of the job — specifically those related to patient interaction. Studies revealed the same trend across multiple other jobs, too, and Brynjolfsson concludes that AI will complement human effort, not replace it.
Put simply, the role of AI in the workplace isn't set in stone. In the same way that staff expectations and employer requirements are evolving over time, so too are the applications of AI.
What can be said for certain, however, is that AI fits the bill for the newest iteration of a GPT. So, while it's impossible to predict exact use cases for AI in the coming years, there's no doubt that AI will change the future of work.
Brynjolfsson's keynote session from the Insights in Action virtual summit and other expert panels on AI's role in the workplace of tomorrow, are available to access on demand.
