California to kick off first open enrollment season with automated system
Connecting state and local government leaders
The state piloted using artificial intelligence to help with automatic enrollment. Now this year’s open enrollment period is underway, the new system faces its first test.
Nov. 1 means the start of another open enrollment period through state and federal health care marketplaces, and for California, this year is big, as the state will rely on its automated enrollment system for the first time.
Covered California — the state’s health insurance marketplace established after Congress passed the Patient Protection and Affordable Care Act — piloted using artificial intelligence in verifying residents’ documents last year. That pilot saw between 80% and 96% of documents verified automatically, meaning a process that could take applicants days or weeks was done in a matter of seconds. It then deployed fully in May, backed by Google Cloud and Deloitte.
But the next few months of open enrollment and plan renewal will be the new system’s biggest test yet, as most of Covered California’s transactions happen during that three-month period, said Kevin Cornish, the organization’s chief information officer. And while technologists may still have nightmares about the bug-riddled 2013 rollout and crash of the federal Healthcare.gov marketplace, using the cloud should be more reliable.
It can also be scaled up and down to respond to demand, Cornish said. While the summer is traditionally a quieter period for Covered California, this fall will be busier. Having the ability to tweak its system’s capacity and deal with higher-than-expected demand is key.
“That's the beauty of the cloud: it scales up and scales down again,” Cornish said in an interview at the recent Google Public Sector Summit in Washington, D.C. He explained that the old system, built in 2014 as a fully on-premises solution, was scaled to meet the demand of the busiest months — November to January — at all times.
“From February to October, it was cruising. It was underutilized. It was a very expensive, wasted asset,” he said.
Covered California typically must verify around 50,000 documents a month and sees many more than that during open enrollment. Under the old system of manual verification, that meant a lot of work for the system’s employees, who had to verify those documents by hand. And it meant uncertainty for potential clients, who would submit their documents for verification and then sometimes not even receive an acknowledgement that they had been received for three days.
The legacy system, which Californians could upload their documents to electronically or send via fax, could only automatically verify 18% of those documents. That left a lot of manual verification.
“I can't imagine a human being that would think, ‘Hey, hooray, it's Monday morning. I'm going to go to work today, and my job today is to validate 100 documents that consumers uploaded. I feel appreciated and it was meaningful,’” Cornish said.
Those tweaks to employee workflow and the immediate impact on their business processes give agency heads those “aha moments” when they see the good that AI can do, said Chris Hein, Google Public Sector’s director of customer engineering. It may still be early days for integrating the technology, and there are plenty of pitfalls ahead, including how to mitigate potential bias and the need for robust governance. But there appears to be a positive future ahead.
“People are not begging to be on the cutting edge, and I think we all probably agree that's not where government should be spending their time: on the cutting edge of technology,” Hein said in an interview at the summit. “But there are a lot of places where there's enough high value that it is starting to become part of the conversation in a way that I find really exciting. …When we get in, we have a conversation with those folks, and we show them how it can change and revolutionize their processes. Their eyes light up in a way that I've never seen before.”
Covered California isn’t done innovating or experimenting with new technologies, Cornish said. He and fellow state leaders are intrigued by the potential for a conversational chatbot that, with a user’s consent, can proactively connect them using information they have entered once with state services — like unemployment insurance — if they, for example, lose their job and need to access other benefits.
That effort would take away what is commonly upwards of a 45-minute process for someone to fill in numerous forms — which often require repeating information — to access benefits. It also would remove the need for the client to know what they are entitled to, and instead put the onus on the government to connect them with services.
It’s early days, however, and Cornish said there is a long way to go before such things can be a reality, either in California or other states.
“We're not ready to say AI is the thing we put out in front of our consumers and have them talk to it, but we're certainly peeling back the layers to get there at some point,” he said. “I'm really excited to envision somewhere down the road a conversational engagement with a consumer.”
NEXT STORY: TSA to allow mobile driver’s licenses after REAL ID goes into effect