The Impact of AI on Journalism
Journalism is an area we do not usually associate with robots, but AI and human news professionals make a great team. While AI has not yet progressed to a level where it can write Pulitzer prize-winning articles just yet, it still offers a lot of benefits for media organizations. Let’s take a look at some of the ways AI is already used in journalism and how it is helping journalists perform their jobs better.
We already mentioned that AI cannot replace journalists who are covering stories and writing in-depth articles about them, but AI can be used to report breaking news. For example, in 2014 an earthquake struck Southern California, and a robot, called Quakbot, produced a small news story with some of the details from the US Geological Survey. By using AI to supplement human journalists, the LA Times was able to get the basic information out to the public within minutes and later on, human staffers wrote more insightful articles about the earthquake. If we fast forward to 2020, we have lots of news organizations using this sort of technology to write news stories on a daily basis. The BBC has something called the Juicer, Washington Post has Heliograph, Bloomberg has Cyborg and the list can go on and on.
These technologies are powered by Natural Language Generation which takes structured data and converts it into natural language. They analyze all of the graphs, tables, and spreadsheets that can be used as a basis for an article and create a text-based on this information. For all of this to happen, human data annotators need to label all of the important information in a training data set such as key concepts, phrases, and names and the system will learn to recognize it independently in the future. It will then take all of this information and produce a small text of its own, which can be used to report important information quickly without having to designate human resources to do the job.
Could Such Robot Reporters Have Opinions?
In addition to reading the facts about a particular issue, we also want to get the opinions of prominent journalists Martin Wolf, Yuval Harari, and many others as well. However, AI is able to overcome this hurdle as well. In fact, even in 2016, we saw the introduction of IBMs Project Debater which competed against human championship debaters on topics such as whether or not governments should subsidize preschools. Later in 2018, IBM rolled out an improved version of the product which is better able to maintain debates with humans. The same principles could be applied to writing editorial articles. While the systems have not developed to a level where they can dethrone the most popular opinionated journalists, there is a lot of promise in this area especially given the machines’ ability to process large volumes of information.
Spotting Biases and Errors
The Financial Times is one of the leaders in using AI to identify new trends, identify biases in their own writing, and fix previous mistakes. Since they were one of the first media companies to successfully introduce a paywall for their content, a lot of emphases was placed on providing highly accurate information. Their Innovations Editor, John Thornhill, jokingly relayed the old adage that there are two people who are willing to pay for accurate news: investors and spies. The Financial Times uses an AI robot called the “Janebot” to analyze the gender ratio of the faces appearing in the newspaper. They also use AI to spot errors they made. The system tracks all user engagement metrics and monitors their feedback. Think of it as crowdsourcing all of the readers of the Financial Times. If there is a mistake in an article, they will pick it up in minutes and report it to the paper.
Mindy Support is Assisting Researchers Develop Next-Gen AI Products
All of the intelligent robots mentioned above require a lot of data annotation to help them learn to read the data and learn to comprehend words, phrases, and patterns. While it takes a lot of time and effort to simply build such an AI robot, the data annotation process can also be very time-consuming. Mindy Support provides comprehensive data annotation services and can take care of this aspect of the project for you. We are one of the largest BPO providers in Eastern Europe with more than 2,000 employees in six locations all over Ukraine. Regardless of the size of your project, we will be able to assemble a team for you and help you meet deadlines. All of the work will be done correctly the first time allowing your project t stay on schedule and eliminate the added costs of redoing particular work.
Posted by Il’ya Dudkin
October 23rd, 2020 Mindy News Blog
Talk to our experts about your AI/ML projectContact us