Earlier this year, Scott Rasmussen, veteran pollster and political commentator, traveled to Bowling Green, Kentucky, known for his Corvette assembly plant and for being the hometown of Fruit of the Loom. After touring the city, about 105 kilometers north of Nashville, he returned home with a novel idea: use AI to transform the surveys, a notoriously volatile and imprecise discipline he had studied for decades. To carry out the project, he associated with an unexpected partner: Google.
Rasmussen had visited Bowling Green to learn about Jigsaw’s work, a group of Google experts who addresses great social challenges, such as online misinformation. At that time, he collaborated with the local government of the city of Kentucky and the surrounding county in an experiment to boost citizen participation. Jigsaw asked residents to answer questions about the issues that worried them most, from the possible opening of a Dave & Buster’s to the debate about the legalization of marijuana. From there, I used Sensemaker, a Google artificial intelligence tool based on its Gemini language model, to analyze the answers and separate the disagreements from the common points of the residents.
Rasmussen declared a Forbes that I was hit by the results and who saw in them an opportunity to try to repair our Balcanized political discourse. Why not survey the entire country?
The goal is to discover points in common, said Rasmussen, who co-founded ESPN with his father, Bill, in 1979. He argues that the American political population is not 50-50, but rather 10-10-80: the 10% that is conservative Maga is at war with 10% of the extreme left, he said. Wait for this project to highlight the rest. “80% intermediate tries to maintain a low profile and avoid being reached by crossfire.”
The problem with traditional surveys, he said, is that the closed questions allow the author to frame or incline the discussion with answers of himself or not, a dichotomy that reduces the nuances. “When you start asking people in another way, or to address their opinions in another way, you listen to things you never thought about asking,” he said.
The result is an ambitious project: on the occasion of the 250th anniversary of the United States next July, Jigsaw was associated with the Napolitan Institute of Rasmussen, a non -profit organization dedicated to the future of surveys and analysis, in an initiative to use AI similarly to probe the Americans about the future of the country. The project, called We the People (us, the people), according to Google exclusively to Forbes will bring together between five and ten people from each of the 435 electoral districts of the US Congress to answer questions about what it means to be American, the most urgent problems facing the country and the future of the nation.
“We want to use AI to give voice and options to people in the world around them,” he told Forbes Yasmin Green, executive director of Jigsaw and Google veteran with 19 years of experience. “If people do not feel that they have a voice, or that their voice does not care about political leaders, they do not feel they are entitled to vote or decision -making capacity.”
“The ability to get people to respond in their own language, to get them to answer as other people have answered questions, is revolutionary for the industry.”
Scott Rasmussen
The project has a positive unit air, but its implications are of greater reach. AI has a great opportunity to transform political surveys. Today, surveys can be wrong and inaccurate; The famous prediction of Hillary Clinton in the 2016 elections. They are usually classified between campaign surveys (whether or not to vote by a candidate) and public opinion surveys (what do you think on a given topic), but both have a similar format. Surveyers call or send text messages with surveys and expect people to respond, then control demography.
Rasmussen said that it could be more difficult for the prediction of whom someone would vote, but still could transform the organization of campaigns. For example, a candidate for a political position could use indications generated by AI to identify issues of interest and importance for their voters, much more depth than the responses of itself or not of traditional surveys. Subsequently, the politician could design a hyperspecific campaign based on that information. “The ability to get people to respond in their own language, to get them to answer as others have answered the questions, is revolutionary for the industry,” said Rasmussen. “It’s a completely different game.”
Read more: there are great hopes of a fed cut at rates next month: this is what you should know
Google’s plan to use AI to survey Americans about their political opinions
Green also sees a broader potential, and mentions market research and public opinion as applications. “When people create a business from your ideas, that really validates it. So I would love that to happen,” Green said. “Our work is to validate its usefulness.”
The project is inspired by a similar experiment in Taiwan. In 2016, the Government decided to bring debates about policies to an online platform called VTAIWAN, which allowed users to discuss legislative recommendations – like the regulation of the sale of alcohol online – and vote them in favor. “Instead of a traditional survey where people remain isolated, these conversations are generative and deliberative, since they allow people to generate new ideas, new feelings with which others identify themselves,” he declared to Forbes Audrey Tang, who helped implement the project before becoming the Prime Minister of Digital Affairs of Taiwan. Since then, Vtaiwan has facilitated around 20 legislative reviews and, in 2023, received a subsidy of $ 100,000 from OpenAi as part of a program to “finance experiments that establish a democratic process to decide which rules should follow the AI systems”.
“We want to use AI to give people voice and options in the world that surrounds them.”
Yasmin Green
Compared to AI, the survey business is relatively modest. In 2025, the market, which includes Ipsos, Pew, Nielsen and Quinnipiac, grew to 8,930 million dollars, a slight increase compared to 8,7 billion dollars from the previous year, according to the Research and Markets firm. It is expected that constantly increases up to 10,230 million dollars by 2029. Companies such as Pew and Quinnipiac, managed from the University of Quinnipiac in Connecticut, specialize in public opinion surveys, while the survey company that Rasmussen founded in 2003, Rasmussen reports, focuses more on the candidates.
His ties with Rasmussen Reports, from which he separated in 2013, could give an unwanted image of the project due to controversies about the alleged conservative bias of the firm. The company has been accused for a long time of favoring Republican candidates, leaning towards older Americans and generating results that favor conservatives. During the US elections of 2024, the company allegedly shared the results of surveys with Trump campaign officials. When asked about that alleged conservative inclination, Rasmussen tries to distance himself from the firm he founded. “I have had nothing to do with this since I left. They have nothing to do with this project, and it really is not a problem,” he said.
The project arises at a time when AI is already leaving a mark on politics and elections around the world. During the US presidential campaign of 2024, a deep falsification of the voice of President Joe Biden told New Hampshire voters not to go to the state primary. In Indonesia, the Beckar political party used AI to digitally recreate Suharto, a dictator who died in 2008, and thus support party candidates. President Donald Trump has repeatedly published memes and videos generated by AI on his social truth platform. Last month, one shared former president Barack Obama detained in the Oval Office.
Jigsaw and the Neapolitan Institute hope to implement AI more productively. The project will consist of three phases. During the first round, participants will answer questions about freedom and equality. For example, “What does freedom mean to you?” Next, Google’s model will generate follow -up questions to deepen the subject, adopting a socratic perspective, Green explained. In this case, I could ask: “If feeling free means expressing yourself without being judged, could you share an example of an occasion when you felt more limited?” The idea is to start generally and, little by little, go in more detail.
In the second round, the Google AI tool will synthesize the responses of the first round on general issues, debate points and data display. From there, participants will have the opportunity to react and share their reflections. In the third round, Google’s AI will create statements based on the analysis of all the previous answers. Finally, participants will vote if they agree or disagree with those statements to see where they find points in common with others.
The organizers are aware that AI could show bias in their indications or analysis of the answers, or ignore the nuances of very sensitive and personal issues when trying to condense thousands of responses on broader issues. To avoid this, said Green, the company will periodically evaluate the AI and will commit to transparency by fully publishing the responses of each participant. After the conclusion of the project, Google will publish a summary report of its findings and publish in open source all the answers, questions and indications used. The company also plans to inform political leaders, groups of experts and academics about the results.
For now, the project is in a pilot phase, with hundreds of participants instead of thousands, and will officially begin in September. Several details are still being determined or have not been published, such as the specific topics that participants will address. Participants will be recruited by Repdata, an external supplier used by the Neapolitan Institute. Google announced that it will share more details of its finished recruitment strategy when the launch date approaches, but added that it repades “will follow a rigorous protocol according to the standards of the sector.”
Normally, pollsters use a method called “random sampling”, which gives all members of a population the same probability of being included. The voices of the participants pondered to take into account populations with less probability of participating in surveys (such as people without a university degree). For the We the People project, the objective is to collect a “representative” sample of people for each congressional district. Therefore, if a district is 90% white and 80% Republican, a representative set could reflect it, Rasmussen explained.
Green is optimistic about the coming results. “I am sure that it will be developed as the conversation that would be desired if the United States met in one place,” he said.
This article was originally published by Forbes Us.
You may be interested: Situational leadership: How to adapt your way of directing for better results
