WP reported on the strong influence of chatbots on election campaign voters

Chatbots with artificial intelligence have a greater influence on the opinion of voters than traditional election campaigns on television. These are the conclusions of studies published in the journals Science and Nature, which were cited by The Washington Post (WP) on December 4.

It is clarified that in one of the experiments conducted in September 2024, an undecided voter explained to a chatbot why he was leaning towards supporting US Vice President Kamala Harris at that time, rather than former US President Donald Trump in the upcoming elections. During the conversation with the chatbot, the voter began to doubt whether it was worth voting at all. This man was one of more than 2,000 Americans who participated in research aimed at studying the impact of AI chatbots on political preferences.

According to the results of the research, chatbots were able to influence the opinion of one of the 25 participants, forcing them to change their initial position and support one candidate or another. While the influence of traditional television advertising turned out to be less significant. The bot supporting Harris convinced about one in 21 participants who had previously been reluctant to vote for her, while the bot supporting Trump attracted one in 35.

"It's really difficult to change people's opinions about political candidates. <...> Therefore, we were surprised to find that chatbots can actually have such a significant impact," said David J. Smith, professor of computer science at Cornell University and co-author of both studies. Rand.

However, despite the obvious effectiveness, researchers have expressed concerns that such technologies can manipulate voters by providing false information. During the experiments, some chatbots were misleading, presenting fake statements as facts. For example, a bot supporting Trump claimed that his presidency had led to a significant increase in jobs and a rise in the stock market, without mentioning the recession and unemployment crisis caused by the pandemic.

In geometric aggression: The new language model from OpenAI will not make a revolution

How did experts evaluate Sam Altman's GPT-4.5 technology, which was taught to be rude to users

OpenAI has banned the use of its ChatGPT neural network for legal and medical consultations in accordance with the updated service usage policy, which became known on November 1. Consultations in the fields of education, housing, insurance, migration and public services are subject to restrictions. The company also banned the use of services for threats, harassment, propaganda of suicide and eating disorders.

All important news is on the Izvestia channel in the MAX messenger.