AI, machine learning, deep learning and automation are no longer future technologies, they’ve already found traction in the enterprise. Whether it’s to organize data, to uncover trends or to make human’s lives easier, AI can have a positive impact on the enterprise – if we let it.
At least, that was the message at The Future of Work in a World of AI, ML, and Automation panel, held at MIT’s CIO Symposium 2018 conference in May, a one-day event where industry thought leaders and educators gathered to discuss the impact of evolving IT trends.
AI won’t wipe out entire job categories
The Department of Labor created a dataset called ONET, which includes the descriptions for 964 occupations in the US. Each job has a list that includes around 20 to 32 tasks, with a total of over 18,000 tasks that exist in the economy.
Brynjolfsson’s team took this data set and evaluated each skillset to determine which tasks could be done better by AI, and which were performed better by humans. The team found that for plenty of jobs there were always tasks that AI could do better than humans, but there were still plenty of tasks that humans excelled at over AI and machine learning.
“This was the pattern we found in occupation after occupation — that very rarely did machine learning just run the table. In most cases, machine learning was able to do some tasks but not others within a given occupation,” Brynjolfsson says. “That means that most of the jobs in your organization will be partly affected by machine learning, but there'll also be things that the humans need to continue to do.”
This will require coordination to help AI and humans work together, “but very rarely will we just completely wipe out entire job categories,” Brynjolfsson says.
Upskilling the workforce
But what about AI-assisted robots? There, the conclusion is the same. In plenty of instances, robots can help alleviate time-consuming, menial, tedious or even physically-strenuous tasks, without employees losing their jobs. In fact, the most likely scenario would see humans and robots working together, with many robots purpose-built for collaboration, also known as co-bots.
We’re still far off from “artificial general intelligence,” which is the type of automation you might see in a movie, where robots can “outwit people on every dimension,” says Brynjolfsson. But AI can make a big difference in the enterprise — especially with any task with sufficient data that has a “set of inputs that map to a set of outputs.”
Panelist Jason Jackson, an assistant professor in political economy and urban planning at MIT, gave the example of healthcare workers. Automating tasks such as patient transfers and lifts can help alleviate some of the physical burden on workers, Jackson says, while also protecting patients and preventing injury. It’s a task that requires strength and effort on the healthcare practitioner’s side — so in this case, automating one skill won’t replace healthcare workers, it will just help them be more effective while also delivering a safer environment for the workers and patients, he adds.
There’s a similar trend in the automotive and manufacturing industries, according to panelist Elisabeth Reynolds, executive director of MIT’s Work of the Future Task Force. She argues that co-bots are creating more upward growth and opportunities for workers now that they’re free to work on more complex tasks. And, while some industries may face displacement, Reynolds says “it is a small percentage of the growth that we see.”
Ownership of AI and machine learning data
The future of robots and AI in the enterprise isn’t without risk. As with past technologies, businesses need to look ahead at potential risks, problems or roadblocks. One key area of anticipated concern revolves around data.
“Many people thought that social media would usher in this wonderful future of connectedness and community, but in many ways it didn’t. It led to some very dysfunctional outcomes. So how do we manage the risk so that AI and automation won’t have those similar unintended consequences?” says Reynolds.
Toronto, for example, let Google install sensors and other equipment to gather data on a street or in a community to deliver valuable information about the city and to discover new insights into the infrastructure, she says. But who owns that data? Does the data belong to Google, or to Toronto?
“It’s obviously owned by the city as well, but does the city have the resources or capabilities to actually do good with that?” Reynolds asks. These are all important questions businesses will have to ask themselves as they commit to relying on data, especially as AI blends further into the workforce and our lives.
As businesses embrace AI, robots, machine learning and deep learning, they will need a clear strategy for leveraging the technology without creating fear around the displacement of jobs or crossing ethical lines.
“The point here is to affirm and assert that there will be work in the future and our challenge is to make sure that that work is a quality work, that is meaningful work, that is accessible work,” says Reynolds.