How Oklahoma has embraced ‘accountable innovation’
SunChan via Getty Images
By
Chris Teale
The Sooner State has run hard towards AI in its bid for government efficiency, but one of its tech leaders said during a recent conference that transparency and training are critical.
Oklahoma has looked to position itself as a leader on artificial intelligence, being one of the first to release task force recommendations for using the technology, and one of the first to appoint a chief AI officer.
But while it may be tempting to see Oklahoma as charging ahead even as lawmakers file bills providing safeguards for AI, Tai Phan, the state’s chief AI officer who has been in his post since November, said a balance is needed. Modernization is no longer optional, he said, and if governments are to scale their ambitions, be efficient and be trustworthy, they must use AI effectively and safely.
“That becomes the point of departure for what AI means for us,” Phan said during the recent Adobe Government Forum in Washington, D.C. “It really reinvents the definition of that and also raises the bar on what that means for our state. AI for us, moving forward, stands for accountable innovation, two words that traditionally may have not played well together. But as we think about our right moment, right now, as leaders shaping AI… it's really about rethinking that definition and what that means moving forward.”
Oklahoma’s report on AI was notable as it suggested that the state could reduce its federal, state and local government workforce from 21% of its population to around 13%, which is closer to “the ideal percentage,” Gov. Kevin Stitt said in a statement at the time. But so far, some of the state’s biggest moves have been in augmenting its employees’ work and making them more efficient, including by using AI to audit its procurement data, fix errors and provide oversight.
Phan said Oklahoma’s biggest focus so far has been on “amplifying the workforce,” and one way of doing that is ensuring the technology is producing reliable answers and solutions and is being used transparently.
“Just because the information is there doesn't mean that person using it is qualified for that,” Phan said. “Reliability here for us means tooling as well as people coming together, and modernization for us is that game between balancing speed and scale. For us, that inflection point between those two outcomes of our journey is really about responsibility.”
Phan also said the state must be transparent about where its AI models get their data and what they are trained on. AI systems “should be able to explain how, why and where did this source come from,” and having the right guardrails in place can help with that.
“Having proper content authority and trust in what's out there is going to be very fundamentally important, especially now that there's public, ChatGPT-like solutions scraping the internet for those solutions,” Phan said. “It really comes down to brand and reputation. How we show up is a vector of what content we put out there.”
For agency heads in Oklahoma, the focus is on rethinking processes as well as embracing new technology, Phan said. It’s a similar story in other states, which have turned to AI and automation only after working out where their business operations are falling short. And moving forward, leaders must also be provided with the right information so they can “become better shoppers of emerging technology,” Phan said of the procurement process.
“We often find ourselves in situations where our state IT organization comes in on the back end and on the front end of a purchase choice, and it creates all sorts of unintended consequences on the ultimate [return on investment] of that financial decision,” he said.
Phan said Oklahoma leaders feel they are on the right path and will be successful in using AI in state government.
“We're really excited about where we're headed, and I think that we have a great path forward as we're rethinking and reimagining what it means to position AI in the construct of a government product and service,” Phan said.