In its second meeting on artificial intelligence in the workforce, the state Little Hoover Commission on Thursday discussed how to use AI itself to predict workforce changes resulting from the technology.
David Beier, a commissioner of the independent state oversight panel, led the conversation, focusing on the definition of artificial intelligence and how to build projections and resulting legislation around AI.
"This subject of artificial intelligence, machine learning, automation is going to change our lives and the lives of our children. And we all have a responsibility, personally and organizationally, to try and understand the dimensions of what we're talking about and the implications, and these are going to evolve over time," Beier said.
Also in the hearing, a major labor organization weighed in — Service Employees International Union (SEIU) California — to voice caution about protecting the role of people as the tech evolves.
The commission discussed enabling workforce projections as a way to shape the education, especially higher education, of a future workforce. There are quite a few estimates on how many jobs could be changed by AI, but Beier acknowledged that no one knows how many will be affected.
"There was a conscious choice to include our friends from organized labor because if we don't focus on workers and the content and evolution of the quality of the job and understand the evolving nature of work, then we're being short-sighted," Beier said.
Irena Asmundson, chief economist for the California Department of Finance, defined AI as an algorithm or computer process that automates production or interactions.
Sara Flocks, policy coordinator for the California Labor Federation, made a distinction: She proposed that automation assist manual labor, while artificial intelligence is a separate concern.
"I think we're facing a very different kind of technological revolution that we haven't seen before," Flocks said. "This is the ability to do machine learning. Artificial intelligence is where a machine can learn to do things and replace workers in every industry and every occupation and change the nature of jobs."
Several attendees agreed that machine learning is based on data, which allows for predictive analytics, making AI different from automation.
"What happens as a result of the algorithm, it's based on the data coming in," Flocks said. "Automation is based on a set of data ... but AI is a much larger set of data, continuously changing, continuously improving, that's going to create the opportunity for actual machine learning."
That seemed to be the component missing for many involved in the discussion. Several people asserted that projecting the influence of AI on future jobs would be strengthened by using data from large companies that use AI, like Google or Facebook.
"It was interesting to hear that the company response rate to labor surveys in California is significantly lower than in Washington state, which means that Washington has access to a richer set of data to help plan for training, investment and, potentially, policy changes," Anne Neville-Bonilla, director of the California Research Bureau, told Techwire via email after the meeting.
And legislation to create standards around AI is in the works.
Sarah Gessler of the California Department of Human Resources said that AI wasn't yet a part of state departments' strategic plans or business goal, mostly because of a lack of information. But she said she looks forward to bringing information to the planning of a future workforce based on AI.
"Because we have a generation coming into the workforce that has all of these ideas already ... and figuring out how to be their destination employer is really where strategic planning and workforce planning and AI come together, Gessler said.
Tia Orr, director of government affairs for SEIU California, contended that the human component cannot be removed from the buildout or use of AI. Workers on the front line would need to be involved in the planning before designing artificial intelligence and would need to be there to correct any mistakes or interact with the most vulnerable populations, she said. Orr also said that more data was necessary and that the best stats usually come out after a negative event, not pre-emptively.
“Data was front and center at today’s discussion," Neville-Bonilla wrote. "We need better data to help us understand what’s happening in California’s economy, particularly with respect to how AI, automation and other advances in technology are changing the nature and, in some cases, the number of jobs.
"In some areas, we need data that may not be collected today — particularly as it relates to the implementation of AI and what outcomes workers and customers are experiencing. In other cases, companies have the data, but are not sharing it. None of us know exactly what impact AI will have on California’s economy, but that’s in large part because we’re at the very beginning of all of this. As many others today noted, there is an opportunity now to help shape the outcomes beyond a simple measure of efficiency to also include ethical standards for use of data and collaborative design and development.”
