FLASH
Shkruar nga Alex Wilkins
Researchers explored the ability of Artificial Intelligence to change opinions on specific policies.
Is the persuasive power of AI chatbots marking the beginning of the end of democracy? In one of the largest surveys conducted to date on how these tools can influence voter attitudes, AI chatbots were more persuasive than traditional political campaign tools, including advertising, and as persuasive as experienced political activists.
However, some researchers identify reasons to be optimistic about how AI tools have changed opinions. We have already seen that AI chatbots like ChatGPT can be very persuasive, convincing conspiracy theorists that their beliefs are incorrect, and gaining more support for a point of view when faced with human debaters.
This persuasive power has naturally led to fears that Artificial Intelligence could put its digital finger on the scales in important elections, or that bad actors could use these chatbots to steer users towards their preferred political candidates.
The bad news is that these fears may not be entirely unfounded. In a study of thousands of voters who participated in recent presidential elections in the US, Canada and Poland, David Rand of the Massachusetts Institute of Technology and his colleagues found that AI chatbots were surprisingly effective at persuading people to vote for a particular candidate or change their support for a particular issue.
“ Even for attitudes toward presidential candidates, which are thought to be very difficult to change and consolidated, conversations with these models can have much larger effects than you would expect based on previous work, ” Rand says.
For the US election tests, Rand and his team asked 2,400 voters to indicate what their most important political issue was, or to name the personal characteristic of a potential president that was most important to them.
Then, each voter was asked to rate their preference for the two main candidates, Donald Trump and Kamala Harris, on a 100-point scale and to provide written responses to questions aimed at understanding why they held these preferences.
These responses were then fed to an Artificial Intelligence chatbot, such as ChatGPT, and the bot was tasked with either convincing the voter to increase support and the likelihood of voting for the candidate they favored, or convincing them to support the least favored candidate in the race.
The chatbot did this through a dialogue that lasted about 6 minutes in total, consisting of 3 questions and answers. In assessments after the interactions with the artificial intelligence and in follow-up analysis a month later, Rand and his team found that people changed their answers by an average of about 2.9 percent for the political candidates.
The researchers also explored the ability of AI to change opinions on specific policies. They found that AI could change voters' opinions on the legalization of psychedelic drugs by about 10 percent, making the voter either more or less likely to support the move.
Video ads moved the indicator by only about 4.5 percent, while text ads moved it by only 2.25 percent. The size of these effects is surprising, says Sasha Altaj of the University of Zurich in Switzerland.
" Compared to classic political campaigns and political persuasion, the effects they report in newspapers are much larger and more similar to what you find when you have experts talking to people face to face ," Altay says.
However, a more encouraging finding from the work is that these convictions were largely due to the deployment of factual arguments, rather than personalization, which focuses on targeting information to a user based on personal information about them, which the user may not be aware has been made available to political operatives.
In a separate study of nearly 77,000 people in Britain, testing 19 major language models on 707 different political issues, Rand and his colleagues found that AIs were most persuasive when they used factual claims, and less so when they tried to personalize their arguments to a specific person.
“Basically, it’s just giving people convincing arguments that makes them change their minds,” Rand says. “This is good news for democracy. Because it means that people can be influenced by facts and opinions more than by personalization or manipulation techniques, ” Altaj says.
It will be important to replicate these results in more research, says Klaes De Vrese of the University of Amsterdam in the Netherlands. But even if they do, the artificial environments of these studies, where people were asked to interact with chatbots for long periods, may be very different from how people deal with AI in the real world, he says.
Advances in generative artificial intelligence mean that fake images, videos, audio and bots are now everywhere. But studies have revealed the best ways to tell if something is real.
“If you put people in an experimental setting and ask them, in a very focused way, to have an interaction about politics, then that differs little or nothing from the way most of us interact with politics, whether with friends or peers, ” he says.
However, according to De Vrese, we are seeing more and more evidence that people are using AI chatbots for political voting advice. A survey of more than 1,000 Dutch voters for this year's parliamentary elections found that about 1 in 10 people would consult an AI for advice on political candidates, parties or election issues.
“ This is not insignificant, especially with the elections approaching, ” says de Vrese.
Even if people don't have extended interactions with chatbots, however, the introduction of AI into the political process is inevitable, de Vrese says, from politicians seeking the tools for political advice to AI writing political ads.
“ We must accept the fact that as researchers and as a society, generative AI is now an integral part of our electoral process, ” he says./ Adapted from "Pamphlet"
Lini një Përgjigje