Researchers have now found that artificial intelligence can alter conspiracy beliefs, challenging the long-held idea that those who embrace these ideas are very unlikely to change. A number of conspiracy theories include the assertion that moon landings were fabricated, and another reports that microchips have been placed in the Covid-19 vaccines. Such beliefs sometimes have a devastating consequence.

The study, headed by associate professor of American University Dr Thomas Costello, revealed that AI indeed effectively promotes critical thinking and disproves fact-based counterarguments. They conducted experiments involving 2,190 people who already subscribed to conspiracy theories using an AI system called “DebunkBot.”.

Participants shared their conspiracy theory and evidence supporting it with the AI and then engaged in a three-round conversation. Afterward, they rated the truth of their beliefs on a 100-point scale. Those who discussed their conspiracy theory with the AI showed a 20% drop in belief, while the change was minimal for those discussing non-conspiracy topics.

“About one in four people who began the experiment believing a conspiracy theory came out the other end without that belief,” said Dr Costello.

“In most cases, the AI can only chip away-making people a bit more sceptical and uncertain-but a select few were disabused of their conspiracy entirely.”

Effects lasted at least two months and applied to almost all types of conspiracy theories, except for those based on factual information. One in four participants who initially believed in a conspiracy theory completely abandoned it by the end of the experiment.

This would suggest that AI may play a pivotal role in warding off the scourge of misinformation, especially on social media. However, Professor Sander van der Linden of the University of Cambridge and others posed questions over whether people would eagerly participate with AI in real-life situations and, quite frankly, how they really managed to win over participants with their methods, including empathy and affirmation. At any rate, this study has worthwhile lessons to impart upon us about the capacity of AI when it comes to countering conspiracy theories.

Sydney:

For more than a decade, researchers and journalists have relied on a digital tool called CrowdTangle to track and fight the spread of viral falsehoods online.

But earlier this week, the owner of CrowdTangle, Meta, shut the tool down. The tech giant has replaced it with its new Content Library, which it says will serve the same purpose and be “more user friendly”.

As long-time users of CrowdTangle to track and analyse online misinformation campaigns, we are sceptical of this claim. We are also very concerned by the fact CrowdTangle’s closure comes at a time when misinformation on social media is rife – and is bound to worsen in the lead up to the November presidential election in the United States.

So, why did Meta decide to close CrowdTangle now? And how will this affect the fight against the spread of misinformation and disinformation online?

CrowdTangle: a tried and trusted tool

Founded in 2011, CrowdTangle quickly gained popularity among media outlets which used it to track and analyse trending topics and articles on social media platforms for commercial purposes. Facebook (as Meta was then called) purchased the company five years later.

There were several reasons why CrowdTangle was a powerful tool for researchers like us who study the spread of misinformation and disinformation online. Firstly, we could download large datasets to use for computational modelling and other forms of analysis. The data was also easily searchable using keywords.

We could also use CrowdTangle to automatically analyse trends in large collections of data from multiple sources.

These features helped us analyse the influence of far-right accounts on Facebook during the COVID pandemic. They also helped us analyse the effectiveness (or lack thereof) of Meta’s content moderation policies in curbing the spread of vaccine misinformation.

But CrowdTangle was not a vital tool only for researchers like us.

Journalists at the New York Times used it to expose the influence of far-right accounts in the US. Their peers at the Australian Broadcasting Corporation also used it to monitor QAnon’s growing influence during the pandemic, and to track disinformation fuelling the recent race riots in the United Kingdom.

Why is Meta closing CrowdTangle?

Given researchers and journalists used CrowdTangle to expose some of Meta’s failings to the world, it’s somewhat unsurprising the company would question why it was funding this largely free service. Keeping it operational, logic would dictate, would be bad for business.

But the reasons behind the closure of CrowdTangle are also more complicated.

One of the co-founders of the company, Brandon Silverman, told Wired the closure was likely part of Meta’s broader withdrawal from the news business.

This withdrawal started in 2016 when the Cambridge Analytica scandal revealed news was bad for Meta’s core business: selling ads.

While news might generate strong engagement on Meta’s platforms, it is the wrong kind of engagement for advertisers. X (formerly Twitter) also recently discovered this, when technology giant IBM pulled its advertising from the platform after its ads were placed next to posts by Nazis.

Meta’s policies aimed at reducing exposure to news content across its platforms make explicit its withdrawal from the news business.

In February, the company announced it would stop “proactively recommending political content from accounts you don’t follow” on Instagram. It also started de-ranking political content on Facebook.

These announcements followed Meta’s decision to pull out of Australia’s News Media Bargaining Code and stop paying for news content.

However, while this redirection away from news may be commercially motivated, it is also about the particular news and information that spreads on Meta’s products (for example, far-right influencers outperforming legacy news).

This brings Meta into the cross-hairs of international regulators and threatens its ability to operate in key markets.

An inferior new tool

Meta has replaced CrowdTangle with its Content Library, which it claims will “provide researchers access to more publicly-available content across Facebook and Instagram”. The new tool will allow access to comments and short-form video, which CrowdTangle did not.

But the new tool has major limitations.

Earlier this month the Coalition for Independent Technology Research published results of a survey with 400 independent researchers and 50 research organisations. They indicated Meta’s Content Library has reduced functionality, including an inability to export data and use external tools to analyse it. It also has reduced access to posts from public figures without a large number of followers.

Combined, these factors will greatly hinder research.

But the biggest issue with Meta’s new Content Library is that it won’t be freely available to journalists and newsrooms.

The exact reasons for this are unclear. What is clear, however, are the implications of this reduced access.

Mozilla, a not-for-profit organisation dedicated to making the internet free and open for all, said in an open letter to Meta:

Meta’s decision will effectively prohibit the outside world, including election integrity experts, from seeing what’s happening on Facebook and Instagram — during the biggest election year on record. This means almost all outside efforts to identify and prevent political disinformation, incitements to violence, and online harassment of women and minorities will be silenced.

There have been countless research reports, news articles, inquiries and lawsuits regarding misuse of Meta’s platforms.

Yet it seems the only lesson the company has learned is to be selective about which data it makes transparent – and to whom.

(Authors:Amelia Johns, Associate Professor, Digital and Social Media, School of Communication, University of Technology Sydney; Francesco Bailo, Lecturer in Data Analytics in the Social Sciences, University of Sydney, and Marian-Andrei Rizoiu, Associate Professor in Behavioral Data Science, University of Technology Sydney)

(Disclosure Statement: Amelia Johns receives funding from the Australia Research Council and the Defence Innovation Network. She has previously received funding from Meta’s content policy award. Francesco Bailo receives funds from the Defence Innovation Network and has previously received a research grant from Facebook (now Meta). Marian-Andrei Rizoiu receives funding from the Australian Department of Home Affairs, the Defence Science and Technology Group, the Defence Innovation Network and the Australian Academy of Science)

This article is republished from The Conversation under a Creative Commons license. Read the original article.
 

(Except for the headline, this story has not been edited by NDTV staff and is published from a syndicated feed.)