Denver Mayor Mike Johnston wants city government to work more like DoorDash.
Need to make a noise complaint? Get a construction permit? Fix a pot hole? The city should be able to respond with the same effectiveness as the app-based food delivery service, Johnston says.
The mayor explained how he hopes artificial intelligence could help the city accomplish that last week while preparing to step on stage at the second annual DenAI Summit. The city-organized conference brought together government workers and leaders of artificial intelligence companies to discuss ways to implement the technology in the public sector.
“For me, the hope is (that) doing this well would actually restore people’s trust in government because we do it better — because we do it efficiently,” he said.
Johnston and his chief AI and information officer, Suma Nallapati — whose role was just expanded to include an AI focus — are planning to make artificial intelligence a major part of how Denver operates. They believe the technology can make government services faster and eliminate rote tasks that eat up their employees’ time.
But it’s not hard to find commentary, from Substack essays to newspaper op-eds, written by artificial intelligence naysayers. Experts warn of AI mistakes, internalized biases, environmental damage caused by the data centers powering AI and threats to jobs as reasons for AI users to proceed with caution.
Scholars tend to suggest that governments should still use the technology — which is becoming increasingly popular in the private sector — but that they must recognize their heightened responsibility to have ethical, transparent policies.
Some of Denver’s own city employees are among the skeptics.
Chris Callanan, a community engagement specialist with the Denver Department of Public Transportation and Infrastructure, said integrating private companies’ AI services into government could have the exact opposite effect on resident trust from what Johnston hopes.
“Is efficiency what the public is looking for? Or are they looking for public value?” said Callanan, who emphasized that he was offering personal views. “I think a lot of the public trust in government that has eroded over the past 40 to 50 years is almost a direct result of the trend of privatization.”
Callanan said government creates value for residents when it creates something new. AI, which generates data based on existing information, is inherently constrained in its ability to do that.
The city has already brought AI into some of its functions, including by creating “Sunny,” an online AI chatbot, to answer residents’ questions. It has used AI to speed up permitting times and improve internal tech operations. Johnston’s administration has also installed more than 100 AI-powered cameras that can read license plates to help law enforcement solve crimes.
Next, city officials plan to solicit more ideas from employees for ways to use the technology. Johnston has said during recent budget discussions that he is considering automating traffic and parking enforcement, an upgrade that could use cameras equipped with artificial intelligence.
In March, the city put out a request for proposals from AI vendors as they seek a broader array of potential applications. Officials are still evaluating the interested companies.
The move toward AI coincides with a tight city budget for Denver. In May, Johnston announced a projected $250 million budget shortfall between this year and next as the city’s costs have gone up and revenues have stagnated.
In August, Johnston’s administration laid off 169 employees and slashed hundreds of vacant positions. He has outlined a number of other cuts to services. The city’s “Sunny” chatbot will take over any resident questions after 5 p.m., for instance, allowing the city’s 311 call center to close two hours earlier.
Where Johnston draws the line
Johnston moderated three panels at the conference with titles like “Reimagining Civic Trust and Empathy with AI” and “The Architecture of Reform: What It Takes to Modernize Government.”
He also asked questions of guests and panelists that hinted at his own views:
“How do we move some of our thinking in the public sector toward higher risk tolerance and higher willingness to test and learn?”
“What are the ways that you see that we can build empathy into these systems and can build connectivity, to make people feel more human in a world that might feel more technological?”
When asked which types of government work might benefit from AI, Johnston didn’t list specific services. He said that generally, any tasks that can be made quicker for employees or more convenient for residents are on the table. But he drew a line, too, when he said sentences for criminals and outcomes of parole hearings are examples of decisions that shouldn’t be made using AI.
When Johnston was elected in 2023, one of his biggest supporters, an independent expenditure committee called Advancing Denver, was funded largely by Reid Hoffman, the billionaire founder of LinkedIn who also is an advocate for AI. Johnston has said that there has been no quid pro quo with his donors.
Nallapati, the city’s chief AI officer, said she still wants employees to verify any information that an AI bot produces.
“AI will never replace human intelligence in my mind,” she said. “Not in our lifetime.”
AI chatbots like ChatGPT frequently hallucinate information and struggle to course-correct. The technology has also generated content that’s “skewed or misleading,” according to the Massachusetts Institute of Technology.
“They’ve been shown to produce images and text that perpetuate biases related to gender, race, political affiliation and more,” according to a blog post from MIT’s Sloan School of Management.
Nallapati said it’s part of her duty to raise awareness about how any AI system’s algorithm works so that any mistakes or bias can be detected early.
“It’s important for us to increase the literacy and awareness of what AI is actually doing,” she said. “Without that knowledge, we are literally letting some company take control of our lives.”
Potential unintended consequences
In 2024, the Partnership for Public Service and the IBM Center for the Business of Government organized a roundtable conversation on AI use in the federal government. The group concluded that the technology had the potential to “make big tasks smaller and onerous ones simple,” but it found AI could also have harmful unintended consequences.
The attendees emphasized that the government needs extensive resources and well-equipped employees to fully address the challenges of ethically implementing AI.
“Government can’t move fast and break things,” says a summary of the event produced by the Partnership for Public Service, referring to a popular motto in the tech world. “It must instead hold itself and its partners accountable for the responsible implementation of AI that improves services and protects the public.”
While Johnston and Nallapati say they don’t see the bots as a way to replace human jobs, Callanan said he and his colleagues often worry about the technology taking their jobs.
“I think that there is really strong potential for technology to be used in a way that hollows out our government institutions,” he said.
While Callanan believes there are obvious places for AI to improve his work, he hopes the city will make its AI use very clear so that residents understand where it’s being used.
Some cities, like Seattle and San Jose, have webpages dedicated to explaining their AI policies. Denver doesn’t yet have a site like that.
Nallapati expressed hope in an interview that instead of fearing AI, Denver could embrace it while creating guardrails.
“Whether we like it or not, the train has left the station,” she said. “So how do we ensure that now we take advantage of the good things and be aware of the bad things?”
Stay up-to-date with Colorado Politics by signing up for our weekly newsletter, The Spot.