Pollsters, the experts who conduct surveys to gauge public opinion on political candidates, have faced increasing scrutiny in recent years. Their accuracy has come under question, especially after major election upsets like Donald Trump's 2016 victory over Hillary Clinton. In the 2020 presidential election, pollsters overestimated Joe Biden's margin of victory over Trump by the largest amount in 40 years. This pattern continued in 2024, when they underestimated Trump’s support for the third consecutive election, leading to criticism from the media and the public. In August 2026, polling inaccuracies were highlighted in key races, including the Democratic Senate primary in Michigan and the Wisconsin gubernatorial race. Polls showed large leads for candidates that did not materialize. In some cases, fake poll data was created by a 21-year-old who set up a website called Median Strategies to show how misleading survey results can be. One of these fabricated polls falsely claimed Los Angeles Mayor Karen Bass was leading in her re-election bid. The poll was even shared by the mayor’s campaign and reported by major media outlets before being exposed as a hoax. Prediction markets, where people bet on election outcomes, have also gained attention as an alternative to traditional polling. Platforms like Kalshi and Polymarket allow users to place bets based on their predictions of election results. These markets have become more popular as poll accuracy has declined. Polymarket’s founder, Shayne Coplan, argues that prediction markets are more accurate than traditional polls. However, data journalist Nate Silver remains skeptical, pointing out that prediction markets can be influenced by factors other than public opinion. In the 2024 presidential election, prediction markets favored Donald Trump, while polls gave Vice President Kamala Harris a slight edge. However, in Wisconsin's recent Democratic gubernatorial primary, prediction markets incorrectly favored a candidate who ultimately lost narrowly. Pollsters also face challenges due to low response rates, making it harder to get a representative sample of voters. To adjust for this, they use a technique called "weighting," which involves adjusting the results based on demographic factors. However, different pollsters use different weighting methods, and these choices can significantly affect the final results. Political scientist Josh Clinton has noted that even small changes in weighting can lead to large differences in poll outcomes. New technologies are also changing the landscape of polling. One emerging method is "silicon sampling," where AI agents simulate human responses to polls. This approach is faster and cheaper than traditional methods and has been tested by organizations like Gallup. However, some experts raise ethical concerns about using AI to simulate human responses. In the 2024 presidential election, AI predictions from a startup called Aaru suggested Kamala Harris would win in four of the seven critical swing states, but she lost all of them. This highlights the ongoing challenges in accurately predicting election outcomes using new and traditional methods alike.