A Kinder, Gentler DOGE? Newsom Says AI Deals Will Make California More Efficient

Speaking in Los Angeles on Tuesday, Gov. Gavin Newsom said a father came up to him during halftime at their children’s basketball game over the weekend and asked him, “Wouldn’t it be good if the blue states would come together and do a DOGE?”

The father was referring to the Trump administration’s Department of Government Efficiency, effectively led by tech billionaire Elon Musk. Newsom said that as he tried to explain that DOGE is not the only example of how to streamline government, nor the first, the man’s eyes glazed over.

“He wants to see the chainsaw,” Newsom said.

In Tuesday’s news conference at the L.A. offices of consulting company Accenture, Newsom announced that the state has entered into three new agreements meant to boost the efficiency of government services — not by mass layoffs but by making use of generative artificial intelligence.

“We’re DOGE, but better … because we’ve been doing it with people, not to people,” Newsom said.

Two of the agreements aim to improve the Department of Transportation’s ability to identify and prevent traffic bottlenecks, accidents and near-misses: one with Accenture to use Azure OpenAI, developed by Microsoft, and the other with Deloitte Consulting to use Google’s Gemini GenAI.

In the third partnership, Department of Tax and Fee Administration officials will use generative AI to build on a pilot project carried out over the last 10 months by SymSoft Solutions, using Anthropic’s Claude, to reduce the time it takes to handle an average customer inquiry.

Newsom said he’s rolling out generative AI projects with multiple corporate partners across eight state departments, at a scale he said has yet to be seen anywhere else in the country. The latest projects build on Newsom’s 2023 executive order directing state agencies to use generative AI technologies to improve state services and help solve intractable issues.

California Transportation Secretary Toks Omishakin said state agencies are committed to delivering on the expectation of better customer service, pointing to the Department of Motor Vehicles as an example.

“Six years ago, there were only a dozen transactions that you could do online. Only 12. Today, there are 50,” he said.

California government’s sclerotic relationship with IT has been the subject of many press inquiries and government reports over the decades. But ethics watchdogs warn generative AI is not necessarily a quick fix, nor an inexpensive one, given that corporate consultants are handling the rollout.

“So much depends on which contexts and how genAI will be used,” wrote Irina Raicu, director of the Internet Ethics program at the Markkula Center for Applied Ethics at Santa Clara University. She argued the state government risks adding to the “pervasive AI hype that’s now endangering, among other things, the technology itself.”

Raicu suggested every state agency should ask itself a set of tough questions ahead of every pilot program: “Why should we integrate generative AI into our processes? Is this the type of AI best suited to the problems that we are hoping to address? Do we have evidence that the AI tool we hope to use works as intended? If it does, is this the most cost-effective way to respond? Have we considered the risks that come with generative AI (and associated issues like its environmental impact), not just the benefits that it would bring?”

While Newsom was talking, the Legislative Analyst’s Office published a preliminary assessment of his generative AI initiative to overhaul the state’s IT project approval process. Upshot: “premature,” without enough information to assess whether the new process would be an improvement on the old.