According to a new study, artificial intelligence (AI) can accurately predict election outcomes and voter behavior.
Phys.org reports that the study examined the accuracy of a GPT-3 language model in simulating the intricate relationship between human ideas, attitudes, and sociocultural contexts of subpopulations.
Brigham Young University (BYU) professors and graduate students in political science and computer science evaluated the AI's accuracy in responding to complex survey questions. What they found is a high similarity between nuanced patterns in human and AI responses.
Simulating the Average Voter
In one experiment, researchers simulated fake people with different traits like race, age, beliefs, and religion. Then they tested if AI would vote as real humans did in the U.S. presidential elections from 2012 to 2020.
They compared the computer program's votes to how real people voted in those elections, using a database called the American National Election Studies. They found that the computer program's votes matched up with real people's votes pretty well. In the study, they call this property "algorithmic fidelity."
What the Results Mean
The study shows that artificial intelligence can help researchers, marketers, as well as pollsters in a number of interesting ways. For example, AI can be used to improve survey questions to make them easier to understand and more inclusive of different people's perspectives.
AI can also create simulations of hard-to-reach groups, like people who live in remote areas or who speak different languages. Additionally, AI can be used to test different survey questions, slogans, and taglines before they are used in focus groups.
Read Also : 'Balenciaga Pope' Francis Storms the Internet, But it Was Only Deepfake using Midjourney AI v5
Ethan Busby, BYU political science professor, and co-author of the study, noted that AI could help people in many different jobs be more efficient.
"We're learning that AI can help us understand people better. It's not replacing humans, but it is helping us study people more effectively. It's about augmenting our ability rather than replacing it. It can help us be more efficient in our work with people by allowing us to pre-test our surveys and our messaging," Busby stated.
You can read more about the study here.
Immediate Concerns
Nonetheless, the rise of artificial intelligence raises a number of questions, including how much it actually knows and how it obtains its data, which people will benefit from this technology, and which populations will suffer as a result of it.
The AI's bias is another issue. Algorithm bias has long been a concern in the rapidly expanding field of artificial intelligence. In one experiment, researchers discovered that flawed AI in robots tends to shape racial and gender stereotypes.
Concerns have also been raised regarding the use of artificial intelligence by con artists and fraudsters to develop increasingly sophisticated phishing scams - especially with the help of more advanced generative AI and deep fake technologies.
While much remains to be determined, the study at hand establishes a set of criteria that researchers can use to evaluate how accurate an AI model is in various subject areas.
Stay posted here at Tech Times.
