As artificial intelligence has permeated texting, writing, shopping and streaming in daily life, government technology officials are wrestling with how to use it responsibly in the public sector.
AI has promise for developing applications that deliver better, faster, and more personalized government service. But artificial intelligence also carries risks. Job displacements, data vulnerabilities, and biased models are all issues that government IT leaders said they must navigate.
At the GovTech Digital Government Summit in Tacoma, senior technology officials at the state, county, and city levels recently explained to Mike McClanahan of TVW’s The Impact, the guardrails guiding their AI strategies.
One sentiment shared by Washington’s Chief Information Officer Bill Kehoe, King County’s Acting CIO Stephen Heard, and Seattle’s Chief Information Security Officer Jake Hammock was that AI should assist, but not replace human beings in decision making and service delivery.
Seattle recently launched an initiative to use AI to improve the permit application process for new residential construction.
“We are optimizing our service delivery. We are increasing the speed of how we engage with our residents,” said Hammock.
According to Hammock, permit applications are a “perfect use case” for testing AI’s potential, but he emphasized, “The City of Seattle’s position is, human in the loop always.”
Heard noted that King County has used machine learning and pattern recognition applications for a long time, but the insights generated by those tools are used to inform human decision making.
“Nothing flows completely through an AI pipeline and then has something happen at the end of it without it being evaluated by a human and I don’t see that changing for us anytime soon,” said Heard.
Kehoe noted that software is already involved in government decision making, but inserting AI into the process will require significant vetting, testing, and outreach.
“We have to make those really critical decisions like who is eligible for what service,” he said.
Artificial intelligence depends on access to data. Ensuring the quality and security of that data is another challenge that must be confronted, according to Kehoe.
“We’re trusted with the residents’ data. It’s a huge responsibility that we have in government and we cannot break that trust,” said Kehoe.