, ,

Getting Comfortable With AI

To learn where agencies will take AI — and where AI will take them — GovLoop spoke with Beth Simone Noveck, New Jersey’s State Chief Artificial Intelligence Strategist and Director of The Governance Lab at Northeastern University.

“For 2024, I think we’re going to see a lot more use of generative AI in government,” Noveck said. “States are going to be more proactive in starting to issue policies of the kind [New Jersey Chief Technology Officer Christopher Rein issued last year] to embrace the use of responsible AI and guide their employees in how to use [it].”

States will clarify how existing privacy and nondiscrimination statutes apply to AI and consider new laws to protect the public’s rights, Noveck predicted.

We Will Learn to Use AI Responsibly

“We’ll see a big push in 2024 to train publicsector workers in responsible uses of these tools that are careful and secure, to ensure that we’re not violating anybody’s privacy or putting up information that’s inaccurate,” Noveck said.

She pointed to New Jersey’s new guidance on state use of AI. “First of all, the policy is very clear about not using personally identifiable information [with] these tools. Second, [it’s about] ensuring that the use of these tools is disclosed. Third, when you ask ChatGPT to write something and are putting it on a website, [you need to review it] in the same way you would if you assigned an intern to help you write something,” said Noveck.

The guidelines also call for ongoing training in the proper use of AI.

The Governance Lab offers workshops on AI in government on its innovate.us website, she noted.

AI Technology Will Mature

Another trend is the increasing maturity of AI technologies, especially in the areas that matter most to governments. “I’m especially excited about the ability to ask generative AI to answer a question based only on a limited set of documents,” Noveck said. “You can feed it with … all the policy documents relating to service X or all the directions relating to how to apply it to service Y, and [it will] provide answers based only on that information.”

“The ability to train on [a specific] set of documents helps to reduce a lot of problems we did worry about in the past and can help ensure greater accuracy and greater privacy,” she added.

AI’s growing multimodal capabilities also are making AI easier to use, Noveck said. Text-tovoice and voice-to-text applications mean AI tools can respond to spoken queries and produce spoken responses far more sophisticated than Alexa or Siri. “I have ChatGPT on my phone, and if I want to, I can go down the street asking it questions. It’s pretty amazing,” she said.

New Jersey has used a platform called Synthesia, an AI prompt-to-video generator, to make training videos. The AI is trained on a human actor’s voice and motions to produce a credible video presentation.

“The other big change is just the integration of AI into the daily way we do things,” Noveck said. It’s moved way beyond autocorrect and autocomplete. For instance, Google Docs “can help you summarize your document or expand your document. Same with the Microsoft. It’s becoming more standard under the hood,” she said.

“It’s one thing to go to ChatGPT and another to be just using your regular Outlook or Gmail and getting a lot of AI along with it,” Noveck said.

This article appeared in our guide, “Gearing Up for AI.” To learn more about AI’s transformative impact in government and prospects for 2024, download the guide here:

Image by rawpixel.com on Freepik

Leave a Comment

Leave a comment

Leave a Reply