Aaron Sanderford
LINCOLN — As artificial intelligence continues to rapidly evolve, some Nebraska lawmakers are viewing a legislative or regulatory role over AI as a balancing act with the First Amendment.
State senators say AI brings opportunities for innovation but also dangers, such as misinformation or disinformation ahead of the 2024 election.
Lawmakers will likely have the opportunity during this year’s session to weigh at least one proposal to regulate AI.
State Sen. John Cavanaugh of Omaha plans to introduce legislation to address certain AI disclosures. His proposal would mirror a Michigan law that requires disclosure of AI if used in political advertisements.
Michigan is the fifth state to enact a law regulating such content, or deepfakes. The others are California, Minnesota, Texas and Washington. More states are considering legislation, as is Congress.
“As technology advances, we need to be sure that it is not abused and take reasonable steps as a state to regulate the use of artificial intelligence in campaigns,” Cavanaugh said in a text.
Unlike the Michigan legislation, which includes criminal penalties, Cavanaugh said his bill would carry civil penalties through the Nebraska Accountability and Disclosure Commission.
‘Wheels of government move slow’
Seven Nebraska lawmakers agreed to review AI-generated clones of their voices as part of a UNL class project. They included State Sen. Beau Ballard, a Lincoln business owner, who said public policymakers are charged with tackling AI, but it won’t be easy.
“The wheels of government move slow,” Ballard said.
Ballard said AI could help with workforce or job development and that it has the potential to be both the most useful and most harmful technology “in probably the history of the world.”
State Sen. Jana Hughes of Seward, an exercise instructor, substitute teacher and former engineering consultant, said lawmakers must consider how to craft bills when the technology “changes tomorrow.”
“You’re chasing something that is exponentially changing,” Hughes said.
State Sen. John Fredrickson of Omaha, a clinical social worker and adjunct professor, said lawmakers must consider these moving pieces and engage with experts and stakeholders.
State Sen. Lynne Walz of Fremont, a real estate agent with a broker’s license and a former teacher, said she’s played around with AI a couple of times and foresees benefits. But she said AI will require voters to do more verification of what lawmakers say.
“The way that our political atmosphere is today, you just don’t put anything past (someone),” Walz said.
‘It’s scary close’
Ballard, Hughes, Fredrickson, Walz and three other lawmakers — Carolyn Bosn of Lincoln, Tom Brandt of Plymouth and Danielle Conrad of Lincoln — agreed to have their voices cloned to see the possibilities.
These “clones” mimicked past legislative speeches by the senators or were used to announce fake resignations, policy positions or other topics of the lawmakers’ choosing.
Hughes said one of her clips sounded exactly the way she would say it. Brandt, a farmer, said he would not have known his clip was AI-generated.
“I don’t think you could find anybody in today’s world that would know the difference,” Brandt said.
Conrad, an attorney and former executive director of the ACLU of Nebraska, said her cloned voice was “shocking” to her. She said the clone sounded a “bit more firm or resolute,” without certain pauses or breaks she tends to use. It was “far more advanced and nuanced” than she would have thought, she said.
This innovation, Hughes said, further complicates the ability to identify misinformation and disinformation.
“It’s scary close, and it’s only going to get better,” Hughes said.
Nebraska will soon join nearly every other state in offering an online video archive of legislative proceedings. The archive will be created once closed captioning of such videos is available, or no later than the start of 2025. The Legislature’s Executive Board will determine the proper use of the recordings.
Nebraska Public Media already livestreams legislative floor debates and committee hearings.
‘A flavor of my voice’
Ballard, Bosn and Hughes reviewed their audio clips as a group. They said cadences, speed, inflections and the “feel” of a voice (robotic or generated) helped clue them into whether it was authentic.
Fredrickson said his cloned voice flow differed from his actual speech, though people are biologically wired to pick up on authenticity. It’s one area, he added, that AI has not yet “finessed.”
“It has a flavor of my voice, but it also doesn’t sound like me at all,” Fredrickson said. “It does sound like a Siri me.”
While the senators may know the sound of their own or their colleagues’ voices, they noted that constituents may not know. All seven said cloned material could confuse voters.
“For a constituent who has never heard me talk, for all they know my voice could sound like Adele’s,” Fredrickson said.
Brandt added that he has heard colleagues on the radio from time to time and thought the audio didn’t sound like them. As AI or similar tech progresses, he said, everyone may begin to doubt what they hear.
‘We have to use it as a tool
’Bosn, a stay-at-home mom and former deputy county attorney in Lancaster and Saunders Counties, said she’s wondered how AI could mpact her children. She has pondered safety parameters, such as in education, to prevent kids from looking up something they shouldn’t.
Walz, who taught fourth and fifth graders, also expressed concern that students could lose an emotional or human connection if they use AI too often in their education.
“I would never want my kids to lose that,” Walz said.
The New York Times has sued OpenAI and Microsoft for copyright infringement, further intensifying a challenge in how developers train AI models. The Times is asking for billions of dollars in damages and for companies to destroy any chatbot models or training data that use copyrighted material from The Times.
Around the country, some states have also looked to define AI in place of regulations. For example, North Dakota lawmakers legislated that “personhood” does not include AI.
Regulation can get tricky with free speech, though, Bosn said, as lawmakers want to advocate for and encourage speech while fostering educational opportunities for children to understand and grow.
“We have to use it as a tool, not a weapon,” Bosn said.
What is ‘proper’ use?
In elections, Ballard said research shows negative campaigning can be more effective than positive campaigning. He and Bosn are running for election after being appointed in 2023.
Bosn said negative campaigning could be met with skepticism or caution among voters, which could help, but she questioned whether lawmakers want to be deciding what is proper.
Hughes, responding to Bosn, said she doesn’t know about all cases, “But doing a fake speech with my voice is improper.”
Bosn said the legal process for holding people accountable can be tricky because it’s time consuming and requires due process.
Past a ‘tipping point’
At the University of Nebraska at Omaha, faculty have begun diving into AI use, including Victor Winter, Ph.D., a professor of computer science, who said recent advances in machine learning and underlying tech infrastructure have led to a “tipping point.”
Development is moving much faster, Winter explained, but AI use is not scary alone. It’s scary for what it could allow humans to do to others, Winter said, quoting Hank Green, an American YouTuber who is half of the team for the educational “Crash Course” YouTube channel.
Cassie Mallette, an instructional designer at UNO, said comfort with AI comes from understanding it, such as experimenting. Mallette said she’s had students who were completely against AI — some who said it “freaked them out” — who later saw how it could be useful in some cases.
“It shouldn’t replace everything we do but should be used as a supplemental tool we can use,” Mallette said in a statement.
Jaci Lindburg, Ph.D., associate vice chancellor of innovative and learning-centric initiatives at UNO, said new tools and technology open up possibilities for exploration, research, learning, training and productivity, with more on the horizon.
“While there will always be some who may use such technologies in a dishonest manner, we are far more excited about the opportunities they create,” Lindburg said in an email.
Finding the ‘right equilibrium’
Brandt said AI will impact all industries and mirror other technological advancements.
For example, when he started farming, Brandt said, he used a six-row planter with markers that someone dropped and had to follow, visually, as best they could. He now has a 16-row planter that is driven automatically to sub-inch accuracy and communicates with satellites.
“It’s an amplification of the path we’ve been on ever since we started the computer age, and we don’t know where it’s going to take us,” Brandt said of AI.
Conrad cautioned that political speech is entitled to the “highest protection” in law and includes speech that is distasteful or hard to hear. She said lawmakers must proceed “very cautiously and very skeptically” before rushing in and crafting laws that could “chill speech.”
It’s all about the “right equilibrium,” she said.
“Popular speech doesn’t need protection,” Conrad said. “Unpopular speech does.”
Editor’s note: This article is based on information gathered for a graduate-level journalism course at the University of Nebraska-Lincoln. The author worked last year as a reporting intern for the Nebraska Examiner.