Have an account?

  •   Personalized content
  •   Your products and support

Need an account?

Create an account

The future of AI in the workforce: Shedding tasks vs. jobs

AI in the workforce is often discussed in light of displacing human jobs. But AI is about displacing certain tasks, not jobs.

As artificial intelligence and robotics wend their way into businesses, there has been some hand-wringing about corresponding job losses.

But those working on the ground in robotics, for example, say that human work will change—not disappear.

“Robots today are taking tasks, not jobs,” said Melonee Wise, CEO of Fetch Robotics—a San Jose, Calif.-based maker of autonomous mobile robots— during a panel discussion on the impact of AI and robotics in the workforce at Cisco Live 2019.

Robots today are taking tasks, not jobs.

Melonee Wise, CEO, Fetch Robotics

Fetch Robotics creates robots that perform repetitive tasks such as picking up and dropping off materials on the manufacturing floor, all while operating alongside human workers. As a result, human workers aren’t bogged down with time-consuming menial tasks and can devote time to higher-level, value-added work.

Still, discussions about AI in the workforce invariably prompt questions about the future of human jobs.

According to estimates, human job losses will flow from the entrenchment of artificial intelligence (AI) in the workforce. Other technologies will also play a role, including high-speed internet connections, cloud computing, and data analytics and robotics.

The McKinsey Institute’s report “Jobs lost, jobs gained,” for example, estimates that “between 400 million and 800 million individuals could be displaced by automation and need to find new jobs by 2030 around the world.” It cites a lesser number of 75 million to 375 million people by 2030 that may need to switch occupational categories and learn new skills.

Roboticists such as Wise say that the field has years to fully develop, but that today AI in the workforce and robotics center on making human work more valuable.

“Our robots enable customers to have associates do the tasks that are high value and get rid of tasks that people don’t like to do and don’t provide as much value,” she noted.

The limits of AI and robotics in the workforce

Other industry experts agree that AI and robotics in the workforce bring opportunities that we must brace for. But they emphasize AI intelligence isn’t yet approaching human intelligence.

“On the one hand, robots are much smarter than us,” said Kate Darling, a research specialist at the MIT Media Lab, in a session on robot-human interactions at Cisco Live. “Robots can work tirelessly . . . and can beat us in things like chess and Jeopardy!. But on the other, people are still much smarter than robots—a lot smarter.”

Darling noted that the endless comparisons between human and AI intelligence present the wrong focus. These sources of intelligence are complementary, not competing.

“One of the misconceptions is that we are constantly trying to compare robots to humans,” Darling said. “It’s a different type of skill set and intelligence than what humans have. We are trying to build a partner in what we are trying to achieve.”

Figure 1: Deloitte Global Human Capital Trends survey on automation technologies

That’s why Fetch Robotics’ robots working in production today execute discrete tasks, working alongside humans. (See Figure 1 for how enterprises now use various automation technologies.)

“We can [have robots] do complicated things in highly controlled, specific scenarios—like welding a car body,” Fetch Robotics’ Wise said. “It’s highly orchestrated. We help the robot do its job well. But we’re not at the stage where we can put a robot in a room and say, ‘Go learn.’”

The limitations of AI intelligence stem in part from the way the algorithms learn; they need large amounts of data to be trained for even discrete tasks. Training bots with these large volumes of data is time-consuming and requires “supervised learning,” where algorithms are fed data categorized by humans.

“Although newer techniques promise to reduce the amount of data required for training AI algorithms, data-hungry supervised learning remains the most prevalent technique today,” noted a McKinsey report on what AI can do for enterprises. Supervised learning is still time-consuming, costly and error-prone. “Even techniques that aim to minimize the amount of data required still need some data,” the report concluded.

That’s clear when we look at speech-activated devices. Semantic intelligence is still immature, though making leaps and bounds. While consumers can order takeout with a voice-activated device or ask a car to dim its interior lighting, semantic intelligence is still task-specific and tone-deaf to context. As a result, it’s often error-prone.

“We are just so far away from having a device that can communicate with you like a human would,” Darling said. “But if it . . . uses speech, people assume that it is smarter than it really is. It makes more sense to use other interfaces right now,” she said.

As a result, at Fetch Robotics, AI-enabled robots are often designed to respond to commands that don’t involve speech.

“People associate speech with intelligence and we don’t want people to associate deep intelligence with these devices,” Wise explained. “We only use beeps and boops. We don’t use any verbal cue like ‘stop’ and ‘hello.’”

Nonetheless, using AI in the workplace can bring business ROI. Not only can robots eliminate menial work by moving “physical things from point A to point B,” Wise said, but also data analytics provides crucial insights.

“Today, the warehouse is a giant black box . . .in terms of traffic patterns, people patterns, worker safety . . . and customers would like to have more insight into that,” Wise explained. “Robots as part of their daily operation can gather that data, and we can do machine learning on top of that to provide business insights.”

AI in the workforce: Tips for staying ahead of the curve

Fetch Robotics’ robots illuminate the opportunity of AI-enabled devices today.

These robots are designed to sweep away a layer of mundane tasks, augmenting human work with more strategic functions. Thus, the question is less about whether AI in the workforce will take jobs and more about how work roles will change as humans work alongside these AI-enabled devices.

And this human-robot combination may yield new opportunities. The World Economic Forum estimates that the entrance of AI in work will create some 60 million net new jobs—out of a total of 133 million new roles by 2022.

“If [robots] can’t replace us, people will need to work with them,” Darling said.

HR executives share this philosophy as they prepare workers for workplace shifts over the coming years.

If [robots] can’t replace us, people will need to work with them.

Kate Darling, research specialist, MIT Media Lab

“I’m more interested to talk about skills or tasks rather than jobs—that prepares people for where the world is going,” said Fran Katsoudas, executive vice president and chief people officer at Cisco. “Robotics and automation and AI will supplement the work we are doing today.”

Enterprises can take certain steps to make AI in the workplace work. Here are a few, (and read about others in Cisco.com’s key tips for AI in the enterprise podcast):

  1. Understand bias in algorithms and establish broad-based organizational trust in the data. Often, AI algorithms are a black box of sorts, with little transparency into their hidden assumptions. In order for AI to bring opportunity, inherent assumptions must be addressed. Experts recommend a diverse group of stakeholders be brought to the table as AI is integrated into the workplace.
  2. Spread the benefits of AI equally. New technologies such as robotics and machine learning can enhance the digital divide. Companies must address how to bring opportunities to workers through emerging technologies rather than widen the gulf of opportunity between haves and have-nots.
  3. Invest in education and training. AI will require new skills—there’s little doubt about that. But companies can be part of the reskilling of the workforce, with programs that target reskilling. [Editor’s note: Cisco invests heavily in IT education and training through programs such as Networking Academy.]

Cisco’s Katsoudas noted that it’s important for enterprises to manage AI’s entry into the workforce and establish ground rules rather than let it run unchecked. Issues such as data privacy and transparency, diversity and training are key before AI becomes even more pervasive in work streams.

“This is a moment before things get too far along for us to set a really healthy foundation,” Katsoudas emphasized. “We have a little bit of time to ensure that everything we do sets our people up for success down the road.”

For more Cisco news:

Subscribe to our newsletter

Lauren Horwitz is the managing editor of Cisco.com, where she covers the IT infrastructure market and develops content strategy. Previously, Horwitz was a senior executive editor in the Business Applications and Architecture group at TechTarget;, a senior editor at Cutter Consortium, an IT research firm; and an editor at the American Prospect, a political journal. She has received awards from American Society of Business Publication Editors (ASBPE), a min Best of the Web award and the Kimmerling Prize for best graduate paper for her editing work on the journal article "The Fluid Jurisprudence of Israel's Emergency Powers.”