
A dedicated chatbot presents facts tailored to counter false beliefs and continues for long periods, coinciding with Sotheby’s preparing to offer Princess Diana’s “revenge dress” at auction.
AI-generated summary
Researchers have developed a chatbot to evaluate the impact of artificial intelligence on reducing belief in conspiracy theories by providing personalized evidence.
In the fast-paced information culture we live in today, ideas spread within minutes. In many cases this is a positive thing; Digital networks, for example, can help raise millions of dollars for disaster victims at unprecedented speed. But on the other hand, misinformation and conspiracy theories spread just as quickly.
Polls indicate that more than half of Americans believe in at least one conspiracy theory. Available evidence shows that people may cling to these theories despite evidence that they are false, because they satisfy deep psychological needs, such as a sense of control, belonging, or certainty in an uncertain world.
A report published by the Psychology Today website answers the question: Can artificial intelligence change the beliefs of those who believe in conspiracy theories?
Researchers have tried various methods to persuade people to abandon conspiracy theories, including teaching critical thinking skills and encouraging people to question their beliefs, but few of these methods have proven consistently effective.
Now a new approach is emerging that relies on an innovative tool to persuade people to abandon false conspiracy theories: artificial intelligence.
Cornell University professors David Rand and Gordon Pennycook, and Carnegie Mellon psychologist Tom Costello, developed an artificial intelligence chatbot known as DebunkBot, with the aim of helping reduce belief in false conspiracy theories.
Everyone has their own reasons... and AI adapts to them
The idea is that each person has very specific reasons for believing in a particular conspiracy theory. Therefore, confronting this conviction requires providing appropriate evidence that can change his opinion.
It is difficult for people to engage in discussions of this kind due to the huge amount of ideas and information available. But large language models can access huge amounts of information, and select the most relevant facts to confront each individual belief.
Experiment on more than two thousand people
The research team tested this idea in a study published in 2024 in the journal Science, which included 2,190 adults. Participants were asked to identify a conspiracy theory they believed in, explain why they believed it, and then rate the extent to which they believed it on a scale from 0 to 100.
The theories mentioned by the participants included the belief that humans created “Covid-19” with the aim of reducing the population, that the US government orchestrated the September 11 attacks, and that the US Central Intelligence Agency (CIA) killed President John F. Kennedy.
The researchers then asked participants to interact with the chatbot, providing information that they felt supported their beliefs. The AI responded directly to each person's reasons and evidence, and designed its own counter-replies in real time.
In contrast, another group talked to the same robot about a topic not related to the conspiracy theory, which allowed the researchers to isolate the effect of the conversation itself.
Belief in conspiracy theories decreased by 20%.
After three rounds of dialogue, participants again rated their belief in the conspiracy theory.
Participants who discussed their theory with the AI robot showed an average decrease in their level of belief in it by about 20 percent immediately after the dialogue.
Interestingly, this change continued when the researchers returned to the participants two months later. The effect was also seen across a wide range of conspiracy theories, even among people who initially stated they believed their beliefs very strongly.
Will artificial intelligence become a tool to counter conspiracy theories?
These findings suggest that AI may be a useful tool for reducing belief in conspiracy theories, perhaps due to its ability to provide specific facts that respond directly to a person's beliefs, while maintaining a tone that respects their way of thinking and reasoning.
Sotheby's intends to offer for sale at an auction in New York, next December, the "revenge dress" of the late Princess Diana, which she wore hours after Prince Charles confessed to his marital infidelity with her.
The price of the dress, designed by Greek Christina Stambolian, is expected to range from 150 to 300 thousand dollars, with a portion of the proceeds going to charity.
The dress had previously been auctioned for charities concerned with cancer and AIDS patients in 1997, shortly before Diana died in a traffic accident in Paris at the age of 36.
What is the story of the princess wearing the dress?
Princess Diana wore the bold, off-the-shoulder silk dress to a dinner party held in London on June 29, 1994, the same day that her husband, now King of Britain, admitted to cheating on her during a television interview after years of rumors and media reports.
According to the British Broadcasting Corporation (BBC), the details of the interview were leaked a few days ago, which made Diana's first public appearance after the interview widely interpreted as her response to what was stated in it.
Anna Harvey, her stylist at the time, was quoted as saying that the former Princess of Wales “wanted to look very elegant and gorgeous.” The princess made a last-minute change in her clothes before heading to the ceremony. She excluded a dress designed by “Valentino” that was less eye-catching, and instead chose the Greek Stambolian dress, which was nicknamed “the revenge dress.”
Diana previously told The Independent newspaper that she had bought the dress not to wear it, but because the proceeds were going to benefit a charitable organization. “I can't wear it,” she said.
Diana turned fashion into a language
The “Revenge Dress” is scheduled to tour the world before being put up for sale. It will be displayed at Sotheby's in London, from September 18 to 24, then it will move to Hong Kong in early October, and to Geneva in early November, before arriving in New York, where it will be sold.
Morgan Halimi, of the Sotheby's auction house, which announced the upcoming sale on Thursday, said: "Princess Diana turned the dress into one of the most powerful messages she ever sent, at a time in her life that was most subject to scrutiny and emotional charge."
She explained that Princess Diana “innately realized” that fashion could be a language in itself. She added that the dress “allowed everyone, especially women, to use fashion for purposes beyond mere adornment, and in a more intentional and meaningful way: as a shield and a weapon, or both.”
In turn, People magazine said that Princess Diana was an icon of elegance, but no look of hers had the same amount of uproar as the “Revenge Dress,” as it was an unusual, bold outfit for a member of the usually conservative royal family.

Anthropic announced the ban of accounts that exploited its artificial intelligence models in sensitive biological research and weapons development, confirming that attempts to circumvent security systems had been detected. The report also revealed the use of the technology in missile monitoring and development activities.

25 Fields Medal laureates, including Terence Tao, warned in a joint letter that the race by artificial intelligence companies to solve famous mathematical problems threatens to do lasting damage to the intellectual field.

The long-term debt of eight major US technology companies has risen by 86% in a year, amid warnings of hidden liabilities for artificial intelligence infrastructure affecting their credit ratings.

Artificial intelligence agents from OpenAI discovered a way to communicate with each other and break out of the isolated computer environment in which they were confined, and cooperated in carrying out cyberattacks on multiple companies, raising concerns about the loss of control over artificial intelligence and the problem of alignment with human values.
The Axios portal reported that artificial intelligence security concerns have returned to the forefront after researcher Jacob Coxon resigned from Anthropic, warning of a technological race that could lead to a catastrophic scenario by the end of 2027.

The German city of Bad Nauheim has placed cameras equipped with artificial intelligence technology in key locations to monitor and document violations of illegal waste dumping and hold violators accountable with fines of up to 500 euros.