The 2020 US Election Is Around the Corner, and Fake News Is on the Way
For the purpose of propaganda and profit, the Internet is full of various false and misleading messages, which are called "fake news". How to identify fake news and prevent being deceived has always troubled the public. To this end, a British startup has developed an artificial intelligence algorithm to identify fake news and prevent it from spreading further.
In recent years, the term "fake news" has begun to appear more frequently, especially in the 2016 US election, which pushed "fake news" to the center of public opinion and increased its usage by more than three times.
According to statistics, on Facebook at that time,Fake news favoring Trump received more than 8.3 million interactions, while the corresponding number for real news from mainstream media was more than 7.3 million.
The US election is imminent this year. Affected by the global COVID-19 pandemic, economic recession, regional disputes and other issues, it is conceivable that the election will face more severe and complex challenges of fake news than ever before.
Creating fake election news to feed a city of young people
There are many reasons for creating fake news, some are motivated by bias, some are motivated by political agenda, and some are motivated purely by economic gain.
In 2016, when the US presidential election was in full swing, more than a hundred Trump-related websites suddenly appeared. They all published news supporting Trump and used US website domain names.
However, these websites are not backed by candidates, nor are they supported by candidates, and they are not even located in the United States. Instead, they all come from Veles, a small town in the small European country of North Macedonia.
Veles is a small town with a population of only 45,000. Its economy is extremely depressed, with the average monthly salary being only 350 euros (about 2,600 yuan). In such an economic environment, it is extremely difficult for young people to find jobs.
Since they were unable to find jobs through formal channels, they started to use crooked means.
They found thatPosting inflammatory fake news that is favorable to Trump on the website will cause Trump supporters to actively spread it, bringing huge traffic to the website or account., you can make a lot of money by relying on advertising.
The cost of creating each piece of fake news is 3 euros. A quick-handed young person can earn at least 24 euros a day, and 500-2000 euros a month.
These fake news are not all fabricated.Many fake news stories are based on real news stories. Fake news writers will add to a real event and fabricate misleading and inflammatory information., integrated into a piece of fake news.
Therefore, faced with a variety of fake news, it is not easy for ordinary people to distinguish their authenticity.
Detecting fake news requires some hardcore work
Lyric Jain, an engineer from Mysore, India, felt the significant impact of fake news on the political situation and people's lives, and wanted to try to use technology to improve this situation.
So he founded a company called Logically that specializes in fake news detection, using artificial intelligence technology to solve this problem.
Logically aims to eliminate bias and misinformation using machine learning, natural language processing, and human oversight.
Recently, Logically received 2.77 million euros (approximately RMB 22.01 million) in financing to further develop its fake news detection products before the US election.
Logically It has signed cooperation contracts with the US public sector and social media platforms to provide technical cooperation for the detection of fake news in this year's US election.
What are the strengths of the technology that was successfully used in the Indian general election?
In 2019, the company’s technology was successfully used in India’s general and local elections.
During the 2019 Indian general election, Logically detected more than 2 million unreliable articles and discovered more than 50,000 pieces of fake news.
In terms of technology, Logically continuously collects data from the Internet.Real-time monitoring of over 1 million domain names and major social media platforms.And news sources, including niche platforms, can be added within hours.
The algorithm designed by the team,A three-pronged approach is used from three aspects: network, metadata, and content, using natural language processing and knowledge engineering technology to extract topics and concepts from messages to better understand the content.
In assessing the credibility of news,Source credibility, article credibility and brand safety, and finally mark the messages as highly credible, medium credible, and low credible.
The world has suffered from fake news for a long time. Can AI be competent in the task of governance?
According to a previous report by MIT's AI Lab, fake news spreads on social media six times faster than real news, and is more likely to be forwarded.
Logically's algorithm has been trained on 700,000 articles and about 20 million sentences.
The algorithm not only filters out fake news, but also highlights the basis for judging it - such as discordant headlines or article content, logical fallacies, political bias and inaccurate statistical information.
In addition, Logically uses a variety of social media intelligence tools on social networks to detect and monitor automated (bot) accounts and other false user information.
The company's fact-checking business involves verifying users' information and ensuring that the views expressed in the messages are based on facts.
Additionally, the company develops custom datasets for fake news, conspiracies, hate speech, and false claims.These basic datasets for domain-specific training are based on standard Natural Language Processing (NLP) problems such as sentiment, event extraction, and NERD (Named Entity Recognition and Disambiguation).
Human-machine collaboration takes off the mask of "fake news"
Founded more than three years ago, Logically now has more than 70 members, but they are not just developers such as backend developers and data scientists.Psychologists, behavioral scientists, fact-checkers, in-house editors and journalists,It is also an important part of the team.
The company has a team of dedicated fact-checkers who use a combination of artificial intelligence and digital forensics to analyze media reports and conspiracy theories to help the public separate facts from lies.
It is also said to be the largest fact-checking team in the world.In addition, there are several in-house journalists who oversee the fact-checkers to ensure that news complies with the International Fact-Checking Network (IFCN) standards.
Users can send their uncertain online messages to the team to check their authenticity
Founder Lyric Jain explains:“We believe AI should complement humans, not replace them. Our technology works alongside news editors and fact-checkers to evaluate news stories."
In just over three months, the United States will be holding a national presidential election. I wonder if Logically can effectively control "fake news" by then and help the American people elect a president who truly reflects public opinion.
References:
-https://www.eu-startups.com/2020/07/fake-news-detection-startup-logically-raises-e2-77-million-to-prepare-for-us-elections/
-https://yourstory.com/2019/08/ai-startup-bengaluru-logically-fake-news-brexit
-https://www.logically.ai/
-- over --