How AI is used in Journalism
07-09-2020 | By Robin Mitchell
The role of AI in everyday life continues to grow, and its applications continue to expand into new sectors and industries. How can AI be used in journalism, what benefits can it provide, and will it see journalists made redundant?
What is AI?
AI is an acronym for Artificial Intelligence and is the process whereby a machine simulates intelligence. Traditional computer tasks are performed by following a set of comparative rules to come to a conclusion, for example, matching a spoken voice pattern to a stored pattern as a form of identification. While AI often uses the same computing fundamentals, it completes tasks using a different approach whereby stimuli are sent through weighted connections to produce a result (this is how neural nets work). Unlike a classical program, such AI systems can be trained to improve their ability to recognise patterns, and are often easier to implement in applications that can carry large amounts of variance. For example, speech recognition is a complex task as people can have different accents, pitches, and speeds of talking. Such a system would be near impossible to program as a large collection of if statements that compare a piece of spoken text to every different variation in how every word in the dictionary can be said. An AI-based speech recognition system is fed lots of different examples of accents, pitches, and speeds, and the resulting network is trained to recognise patterns that exist in all the examples (for example, formants found in the speech).
What problems does modern journalism face?
Ironically, answering this question proves a major issue faced with journalists in general; inherent bias and opinion. No matter the intentions of a writer, almost all journalists inherently contain a level of bias, whether it be political, economic, social, or all of the above. But opinion and sensationalism are also a common problem, and it affects the science and tech fields just as much as it affects fashion and celebrity industries. In many other cases, journalists also suffer from poor sources, or parroting articles that themselves not original, and thus the true meaning from a research paper or product announcement gets lost in a similar way to “Chinese Whispers”. With so much data available on the internet, and the difficulty in determining fact from fiction, modern journalists are faced with an ever-growing problem of regaining trust from the public.
How AI can help journalism
One of the biggest advantages of AI is the ability to recognise patterns, and such systems can be trained to recognise the difference between genuine reports, and those that are fake. AI also has the ability to be trained by journalists to learn how to piece factual information together, thus allowing for a system that can take daily temperature readings, sports information, and press releases to automatically write readable articles that only report facts. Such systems can also be trained to ignore opinion pieces that heavily use descriptive language, while also simultaneously ranking journalists based on integrity and ability to write opinion free pieces.
As it turns out, AI journalists are already in operation and write thousands of articles daily. One example of an AI journalist system is Quakebot; this AI journalist can write reports on earthquakes three minutes after they have occurred. Data from the Geological Survey is obtained by the bot, and AI is used to create a legible article that reports on the incident. According to Forbes, multiple major news outlets now incorporate AI systems including BBC (called Juicer), The Washington Post (called Heliograf), and Bloomberg (called Cyborg). Such systems take in data, extract important information, and then pass them through a natural language generation software to create the article. However, such articles can only produce content from large amounts of existing information, and as such cannot form opinions or write engaging content.
Will AI remove journalists from the equation?
While it may seem that AI could remove the need for journalists, the truth is that many readers enjoy content containing opinion and bias. Unless an AI system can be made to be biased, or form an opinion, then it is unlikely that AI will replace all journalists in the near future. But from someone who thrives on truth and facts, it is hard to call those who write opinion “journalists”, because opinion itself is not fact, and can be used to spread misinformation (it is hard to find any single news organisation that carries no bias or that prints opinion). Thus, an AI-driven news site that does not post opinion pieces may help to restore integrity to the media industry, and entirely AI news sites may become the initial source that other journalists can work from.
