Using AI tools to fix AI-based product failures

Fake news is getting more sophisticated with AI manipulation

Recently, at the meeting between the Vietnam Journalists Association and the Thai Journalists Association held in Hanoi, the leaders of the two Associations spent most of their time mentioning the problem of "fake news" and stated that this is no longer a national problem but a major problem, causing unpredictable impacts on the region and the world.

According to research by Google DeepMind, the creation of images, videos, and audio that impersonate humans is nearly twice as common as the creation of misinformation using tools like AI chatbots.

Deepfakes of British Prime Minister Rishi Sunak and other global leaders have appeared on TikTok, X and Instagram in recent months, with concerns that despite efforts by social networks to label or remove the content, audiences may not realise it is fake and that its dissemination could influence voters.

Fake artwork of British Prime Minister Rishi Sunak has appeared on TikTok and Instagram ahead of the general election. (Photo: AFP)

Assessing the impact of fake news in the technology era, MSc. Luong Dong Son - Lecturer at the Institute of Journalism and Communication, Academy of Journalism and Propaganda, said that the product of the combination of AI and misinformation, also known as surreal fake news, can distort public perception of events and social issues, thereby affecting the way they assess the situation and make decisions. This can cause serious consequences, from small personal decisions to important decisions affecting the community and the country.

Hyper-fake news can be used as a tool to manipulate public opinion, incite violence, sow division, and undermine social stability. In the political arena, hyper-fake news can erode trust in democratic institutions, influence election outcomes, and even incite conflict.

Regarding the challenges for journalists and news organizations, MSc. Luong Dong Son said that the sophistication of surreal fake news, especially deepfake, makes it increasingly difficult to distinguish between real and fake, even for experienced journalists.

"In the age of information explosion, the public increasingly demands that journalists and news organizations provide information quickly and accurately. However, this pressure can lead to journalists hastily publishing information without carefully verifying it, creating opportunities for surreal fake news to spread. In addition, journalists and news organizations face legal and ethical risks if they are not careful in verifying and publishing information," said Mr. Son.

Get AI to verify information

According to Master Luong Dong Son, to deal with this type of fake news, applying modern technological solutions is extremely necessary.

"These solutions not only focus on detecting and eliminating fake news but also preventing them right from the early stages of production," said MSc. Dong Son.

First, one of the most important technologies today, according to Mr. Son, is using artificial intelligence systems to detect fake news. Modern AI systems are capable of analyzing context, comparing information from multiple sources and detecting abnormalities in information content. A typical example is the use of AI to analyze videos and images to detect signs of editing or production using deepfake technology.

Next, social media platforms and search engines are also actively deploying machine learning algorithms to automatically classify and filter out fake news content before it goes viral. These algorithms are designed to recognize anomalies in sharing patterns, speed of spread, or even the tone of posts, thereby assessing the likelihood that the information is fake.

In addition, the use of blockchain technology to manage the origin of information is also a potential solution. Blockchain has the ability to create a transparent and immutable authentication system, thereby helping to track the origin and history of all information, ensuring accuracy and reliability.

AI can create fake content or misleading information and images that are difficult to distinguish, causing great damage to the reputation of individuals and organizations, and even harming national and ethnic interests.

Sharing the same opinion with MSc. Luong Dong Son, Mr. Phan Van Tu - Head of the Department of Journalism - Faculty of Journalism and Communication, University of Social Sciences and Humanities, Ho Chi Minh City emphasized that, in addition to providing skills to exploit AI's strengths for journalism, we also need to use AI tools to verify information.

Source verification, which is essentially assessing the reliability of a source, is an important skill that involves verifying the source of information, comparing it with other sources, and using fact-checking tools to check reliability.

Interestingly, there are now many AI-integrated tools that support journalists in the verification process, in addition to traditional tools that support fact-checkers such as Google Image, Tineye reverse image search, Whopostedwhat.com, Waybackmachine, Webarchive.org, InVID, WeVerify, ExifTool, Metapicz...

Mr. Phan Van Tu said that training journalists' professional skills in the current period requires focusing on providing information assessment capacity against the attack of fake news manipulated by AI. "Using AI tools to handle AI-generated errors is the fastest way to help journalists and press agencies not only perform their roles well but also contribute to improving the quality of information and protecting the public from false information and negative impacts of the media," said Mr. Tu.

However, Mr. Tu also pointed out that journalists' ability to assess information cannot rely solely on tools. It must be said right away that AI can support the analysis of news content but cannot detect all types of fake news and manipulated information. AI software can still be fooled by the very AI techniques it uses to detect.

"No tool is a magic wand, but AI tools that detect fake photos/videos are still a valuable resource as part of a verification process or due diligence method," said Phan Van Tu.

Hoa Giang

Source: https://www.congluan.vn/tin-gia-trong-thoi-ky-tri-tue-nhan-tao-dung-cac-cong-cu-ai-de-tri-cac-san-pham-loi-tu-ai-post308981.html