Why AI Will Create Jobs, Not Destroy Them
Coverage of the Investment Innovation Conference
BY Caroline Cakebread | July 10, 2018
Dire warnings about job losses caused by the rise of artificial intelligence (AI) in the workplace have sparked a good deal of debate about their use. But are the warnings overblown — and could machine learning actually lead to greater job creation?
During a panel discussion on AI in the workplace, Matthew Killi, head of machine learning and development with Toronto-based Deep Learning, says machine learning won’t replace people – rather, people should see these applications as another part of the team that is able to do things humans can’t, like process massive data sets that would otherwise be indigestible.
“There will be disruption,” he says. “But job creation will be enormous with AI.”
Killi adds, “There is an expectation that AI robots will be so good they will limit man hours – in fact that’s not true.” In fact, AI creates more efficiency, allowing companies to add workers to handle the increased output.
Shannon Vallor agrees that AI can increase jobs and points to Amazon as an example: that company tripled the number of robots in its warehouses and added human workers too. However, while AI can enhance people’s ability to do their job, Vallor warns that we need to understand that it also has its weaknesses: “People get so sold on the power of AI but no one wants to tell them about the technology’s limitations.”
Specifically, there are things AI can do – and can’t: “If the job can’t be mathematically expressed, then the machine can’t do it,” Vallor says, adding that for AI to work “you need data and the right kind of problem.”
A matter of ethics
At the same time, ethical concerns abound: we need to ensure that machine values serve human values and not the other way round warns Vallor. “It doesn’t become worth having just because it’s more efficient or predictable.”
For his part, Killi isn’t optimistic about regulators’ ability to oversee AI. “I think there needs to be a paradigm shift in how we regulate,” he says. “Machine learning is not rule-based. We need a new framework.” If left to the private sector, however, it will be guided by self-serving interests.