Researchers found an AI bot is pretty good at helping people rethink conspiracy theories

Everyone on here would benefit a few minutes asking a chatbot about a topic before posting their 2 bits on here.

"How true or accurate is this? Explain how it could be true? What are the counterfactuals?"
​

Granted, this isn't a permanent solution. As long as there's a vested interest to believe or be distracted by false narratives, there will be a cat-and-mouse escalation. I'm sure red-blooded Xitter dwellers will scoff at a bot like GPT because of TECH ELITE BIAS, but using Grok would also produce similar results.

https://www.washingtonpost.com/opinions/2025/02/26/ai-research-conspiracy-theories/

Michael Protzman, a former Seattle demolition contractor, attracted a following of QAnon conspiracy believers so devoted to his prophecies that hundreds of them traveled to Dallas on Nov. 1, 2021, to witness John F. Kennedy and John F. Kennedy Jr. appear on a grassy knoll adjacent to the assassination site of JFK and ordain Donald Trump “the king of kings.”

When the deceased Kennedys failed to show, Protzman’s supporters followed him to a nearby Rolling Stones concert, where he predicted that Keith Richards, Mick Jagger and Steve Jordan would remove their masks and reveal their true identities: JFK, Michael Jackson and Prince, respectively. When the unmasking didn’t occur, many members of the group waited around in Dallas for months, raising money from online supporters to pay their bills at the Hyatt Regency.

The Dallas fiasco cost Protzman some followers, but it hardened the resolve of the rest. His most devoted believers even refused to believe it when Protzman died in 2023.

The case of Protzman and his QAnon followers is not unique. Researchers have struggled for decades to develop techniques to weaken the grip of conspiracy theories and cult ideology on adherents.

This is why a new paper in the journal Science by Thomas Costello of MIT’s Sloan School of Management, Gordon Pennycook of Cornell University and David Rand, also of Sloan, is so exciting. It finds hope in new technology: a conversation partner powered by artificial intelligence.

In a pair of studies involving more than 2,000 participants, the researchers found a 20 percent reduction in belief in conspiracy theories after participants interacted with a powerful, flexible, personalized GPT-4 Turbo conversation partner. The researchers trained the AI to try to persuade the participants to reduce their belief in conspiracies by refuting the specific evidence the participants provided to support their favored conspiracy theory.

The reduction in belief held across a range of topics, including covid-19 and the outcome of the 2020 presidential election. Even more encouraging, participants demonstrated increased intentions to ignore or unfollow social media accounts promoting the conspiracies, and significantly increased willingness to ignore or argue against other believers in the conspiracy.

And the results appear to be durable, holding up in evaluations 10 days and two months later.

It’s a surprising result, especially given the previous failures of scientific interventions to significantly reduce belief. Taken with the ineffectiveness of real-life events to change minds, as in the Protzman case, and the fact that no amount of arguing at Thanksgiving seems to matter, the picture to date has been bleak.

Why was AI able to persuade people to change their minds? The authors posit that it “simply takes the right evidence,” tailored to the individual, to effect belief change, noting: “From a theoretical perspective, this paints a surprisingly optimistic picture of human reasoning: Conspiratorial rabbit holes may indeed have an exit. Psychological needs and motivations do not inherently blind conspiracists to evidence.”

That conclusion raises a question: Why is it that Kennedy and his son failing to show — evidence that would appear perfectly tailored to the conspiracists — did not loosen these false beliefs, but tailored AI arguments seem to work? I suspect the difference lies in how participants perceived the technology. Given it is nonhuman, participants probably saw it as a neutral source and not a threat to their identity.

When someone takes an extreme position, they’re increasing the distance between themselves and the pack. That distance makes the position more integral to identity, a part of the way that a person defines themself as distinct from other people. Once a belief is integral to identity, it sticks.

Seventy years ago, Leon Festinger, who originated the concept of cognitive dissonance, spent months observing members of a doomsday cult as they awaited the arrival of aliens that would wipe out humanity but spare true believers on Dec. 20, 1954. When no alien spaceship arrived and Earth did not perish in a flood, most of the followers did the same as many of the QAnon believers in Dallas: Counterintuitively, they held fast to their belief in the prophecy and escalated their commitment to the group. Identity trumped facts, even ones perfectly tailored to refute the prophecy.

Our intuition also tells us the right incentives ought to sway people toward reality. Yet not even financial incentives are enough to trump identity. John Beshears, now of Harvard Business School, and Katy Milkman, a professor at the University of Pennsylvania’s Wharton School and author of the 2021 bestseller “How to Change,” examined 18 years of corporate earnings estimates and updates of more than 6,000 stock analysts. They wanted to know whether analysts who made earnings estimates that were way out of the consensus changed their positions when, later, the actual results didn’t bear out their forecasts.

The analysts didn’t change their minds. And it isn’t just that stock analysts are stubborn, because when the same analysts made consensus earnings estimates, they did update their forecasts when actual earnings went against them. It was only when they stood out from the crowd that they held fast to their beliefs.

Stock analysts don’t have a financial incentive for sticking to inaccurate estimates. In fact, it’s quite the opposite: They were punished for their stubbornness when they made incorrect earnings forecasts. In that way, the analysts behaved similarly to cult members who had taken an identity-defining, extreme position.

It is hard to walk away from who you are, whether you are a QAnon believer, a flat-Earther, a truther of any kind or just a stock analyst who has taken a position that makes you stand out from the crowd.

And that’s why the AI approach might work so well. The participants were not interacting with a human, which, I suspect, didn’t trigger identity in the same way, allowing the participants to be more open-minded. Identity is such a huge part of these conspiracy theories in terms of distinctiveness, putting distance between you and other people. When you’re interacting with AI, you’re not arguing with a human being whom you might be standing in opposition to, which could cause you to be less open-minded.

It has been a dispiriting 70 years or so for the science of belief change since Festinger first identified cognitive dissonance. And that’s what makes the promise of AI in reducing belief in conspiracies so exciting.
​

Next steps:

• Enshittify the current free or "cheap" chatbots"
• Sew distrust/dislike of current bots, and repeat claims of bias that would open a market for billionaire Con puppet masters training their own Balanced and Free LLMs for anyone lazy or dumb enough not to make or customize one in their iphone 20s.
• ???
• A cult fanbase 4 Moar Profit