AI in Journalism
AI is continually making waves in the news and journalism industry. The work of communicating everything that happens in the society seems like it's something entirely human-dependent. Furthermore, what does a machine have to do with what happens in the news?
Journalism, as we know, has a large consumer base and is highly dependent on technology from the collection of data from different news sources to the publication of the content.
Despite journalism being one of the consumers of technology, technological advancements change the way the news and journalism are carried out. AI is continuously focused on improving news production, publishing, and sharing ideas in particular.
In automated or algorithmic or robot journalism, computer programs are used to generate news articles. Computers, rather than human reporters, automatically produce stories through AI (Artificial Intelligence) software. These programs organize, interpret, and also present the data in a readable manner.
Artificial intelligence in journalism involves algorithms that scan large amounts of given data, select from pre-programmed articles, order key points and insert information such as places, amounts, statistics, and names. There is also the potential to customize the output to fit a specific tone, style, and voice.
Common Worldwide Data Leaks
In recent years, investigative journalists had a protective duty between the public sphere and their confidential information sources. However, this duty has been undergoing changes in the last two decades with the introduction of new media.
Some instances like Snowden and WikiLeaks show how contemporary the media allows people to directly release data to the global audience, which raises the question of how the operations of journalists are affected by the recent leaks.
- The Bahamian Files
In the Bahamas, some 400,000 documents were leaked from the register of legal entities on offshore companies, including Ukrainians. There was data of about 60 companies on the Ukrainian part of the Bahamian archive. Massive datasets include information on citizens of more than 100 countries who had applied to one of the globally famous offshore zones, the Caribbean jurisdiction for the registration of their businesses.
The Bahamian files were publicly made available on midday. The documents had information on different Germans' offshore secrets. The DDoS (Distributed Denial of Secrets) published a full archive of the documents, unlike the ICIJ, who posted the Bahamas leaks, with no documents. The archive posted by the DDoS contained much larger data amounts than the Bahamas Leaks with some of it, namely until 2018, relating to a very recent period.
As mentioned earlier, the Ukrainian part of the Bahamian archive contained information of about 60 companies. The companies were mostly founded by Ukrainians, low profile businessmen who were not part of the government, but several names drew public attention.
It's not a crime to register companies in offshore jurisdictions. It is a global exercise that occurs for various reasons, such as convenience in the structuring of one's business and quality services in the offshore zones.
- FDA Data Dump
You may have only heard of Panama Leaks or WikiLeaks. However, every few months, there is a leak of information that journalists cover. For example, the FDA data dump revealed breast and dental implant problems among hidden reports. The majority of the reports about the 5.8 medical device problems that the FDA accessed were breast and dental implants.
Many medical device complaints are kept at the FDA, where they are accessible to the general public. The 5.8 million complaints were the 'alternative summary reports' that were hidden for decades, and the FDA only could access the files; nurses and doctors couldn't even see them.
- The Luanda Leaks
Isabel dos Santos, a businesswoman, was accused of making a fortune at the Angola citizens' expense, as revealed in the Luanda Leaks. She moved billions through her shell empire, which was made possible by consultants, lawyers, and accountants.
The businesswoman claimed that her success was self-made, but an investigation, Luanda Leaks, by 36 media partners and the ICIJ revealed the true source of her wealth. Isabel dos Santos in her defense, accused the Luanda authorities of using falsified documents to freeze her assets as presented in her evidence.
The Luanda authorities denied the claims of Isabel alleging false emails and a fake passport. They further stressed that the estimated damages in the proceedings against her are more than 5 billion dollars by the state's estimates.
There have been massive data leaks in recent years from the offshore jurisdictions like the offshore leaks, Panama papers, Bahamas leaks, and paradise papers.
Incorporation of Machine Learning (ML) in Journalism
In some instances, machine learning cannot do something you couldn't; but what machine learning does is accomplish it faster than humans. Machine learning has the potential to find emails, the same as the ones you already have, assist you in finding frames of a given video with a senator, etc. The computer needs to be fed particular knowledge on the questions you are trying to find solutions.
This knowledge helps you solve several common problem patterns such as flagging matching things from given sets of documents, filtering complex data point spreadsheets, searching through picture piles, and sorting caches of reader tips.
It's up to you to determine whether the images, data points, or the documents found via machine learning are newsworthy or exciting. Still, ML can get you from an unmanageable pile to a manageable one. You should also know that the computer can be mistaken, confused, and fail to understand your question like any other source, so continue with your journalism. ML is continuously being incorporated and advanced in the journalism industry so that it perfectly answers questions that reporters have concerning their data.
Machine learning techniques greatly help journalists in daily stories and accomplish some investigations that may involve going through vast piles of documents for months, which can be done in just a week. Journalists have been known to utilize machine learning tools and codes like in the following real-world examples:
- finding assaults classified as minor but were serious ones
- detecting sexual abuse complaints in disciplinary reports
- detecting political advertising
- finding the discussed topics by the members of congress
- detecting toxic comments
Learning ML With Free Hands-on Videos
As a journalist, you can learn how to how to incorporate ML in your investigation tasks through several online coding notebooks and videos. Get to try your hands in searching images, analyzing text with AI, and sorting images.
You can enter these pre-trained machine learning lessons by John Keefe, an editor at Quartz Al Studio, and a member of the Mauritius leaks. There are 15 lessons which are 4-15 minutes. You'll get through a series of project examples focused on journalism like sorting docs into piles, detecting objects in images, and also the extraction of individuals' names from troves of texts.
The lessons utilize the fast.ai ML library for python, which makes Machine Learning easy for individuals who are not well versed in computers and math. The videos can be your preferred journalist oriented primer for the free and excellent fast.ai course practical deep learning for coders, which will significantly help you.