How Artificial Intelligence Helps Detect Fake News Instead of Creating It

2023-06-20

Fake news is a complex issue that can span across text, images, and videos. In particular, there are several methods for generating fake news articles, especially for written articles. Fake news articles can be created by selectively editing facts, including names, dates, or statistics. An article can also be entirely fabricated with fictional events or characters. Fake news articles can also be generated by machines, as advancements in artificial intelligence have made it particularly easy to produce false information. Destructive Impact For questions like "Was there voter fraud during the 2020 US elections?" or "Is climate change a hoax?", fact-checking can be done by analyzing available data. These questions can be answered with "true" or "false," but there is a possibility of potential errors or misleading information surrounding such questions. The spread of false information and fake news can cause harm to many people in a short period of time. While the concept of fake news existed before technological advancements, social media has exacerbated the problem. A 2018 Twitter study showed that fake news is more likely to be retweeted by humans rather than bots, and the likelihood of retweeting fake news is 70% higher compared to true stories. The same study found that it takes about six times longer for true stories to reach a group of 1500 people, and true stories rarely reach more than 1000 people, while popular fake news can spread to as many as 100,000 people. The 2020 US presidential election, COVID-19 vaccines, and climate change have all been targets of misinformation campaigns with severe consequences. It is estimated that misinformation surrounding COVID-19 causes daily losses between $500 million and $3 billion. The cost of political misinformation can be social unrest, violence, and erosion of public trust in democratic institutions. Detecting Misinformation Detecting misinformation can be achieved through the application of algorithms, machine learning models, artificial intelligence, and human analysis. An important question is, once misinformation is detected, who is responsible for controlling or even stopping its spread. Only social media companies truly have the capability to control the spread of information through their networks. One particularly simple yet effective method of generating false information is selectively editing news articles. For example, consider the statement "Ukrainian director and playwright arrested and charged for 'defending terrorism.'" This is achieved by replacing "Russia" with "Ukraine" in the original sentence of a real news article. Multiple methods need to be employed to detect false information on the internet in order to control its growth and spread. Communication in social media can be modeled as a network, where users form nodes in the network model, and communication forms links between nodes; reposting or liking a post reflects a connection between two nodes. In this network model, individuals who propagate false information often form a densely connected core-periphery structure compared to users who propagate the truth. My research team has developed efficient algorithms for detecting dense structures from communication networks. These pieces of information can be further analyzed to detect instances of false information propagation. Since these algorithms rely solely on communication structures, content analysis is required by algorithms and humans to confirm instances of false information. Detecting manipulated articles requires careful analysis. Our research utilizes neural network-based methods that combine textual information with external knowledge bases to detect such manipulation. Preventing Spread Detecting false information is only half the battle; decisive action needs to be taken to prevent its spread. Strategies for combating the spread of false information in social networks include intervention by internet platforms and launching counter-activities to neutralize false news campaigns. Interventions can take the form of strict measures, such as suspending user accounts, or milder measures, such as labeling posts as suspicious. Algorithms and AI-driven networks are not 100% reliable. Incorrectly intervening in genuine content or failing to intervene in false content both come at a cost. To address this, we have designed an intelligent intervention strategy that automatically determines whether to intervene in content based on predicted authenticity and predicted popularity. Fighting Fake News Launching counter-activities to minimize or neutralize the impact of false information propagation requires careful consideration of the major differences in the speed and scope of dissemination between truth and fake news. In addition to these differences, people's reactions vary for stories of different lengths, topics, and users. Our approach takes all these factors into account and designs an efficient counter-activity strategy that effectively mitigates the spread of false information. Recent advancements in generative artificial intelligence, particularly the development of large-scale language models like ChatGPT, have made it easier than ever to create articles at a rapid pace and in large quantities, thereby increasing the challenge of detecting false information and countering its spread in real-time. Our current research is dedicated to addressing this ongoing challenge with significant societal impact. Source: https://gizmodo.com/ai-chatgpt-detector-fake-news-generate-misinformation-1850549536

RECENT AI NEWS

RECENT AI TOOLS