Can Artificial Intelligence Save Journalism?



Artificial intelligence and its impact on journalism

We are at a crossroads. A crossroads that will in large part determine the future of journalism. The Covid-19 pandemic has caused an unprecedented crisis that could decimate certain media organizations.

One possible solution has been proposed: artificial intelligence (AI). AI refers "to intelligent machines that learn from experience and perform tasks like humans," according to Francesco Marconi, a professor of journalism at Columbia University in New York who has just published a book on the subject: Newsmakers, Artificial Intelligence and the Future of Journalism.

Francesco Marconi is not just anyone. He was head of the "media lab" at the Wall Street Journal and the Associated Press (AP), one of the largest news organizations in the world.

His thesis is clear and incontrovertible: the journalism world is not keeping pace with the evolution of new technologies. So newsrooms need to take advantage of what AI can offer and come up with new a business model.

For Marconi, we are missing the boat and AI needs to be at the heart of this future business model. As a professor of journalism at UQAM, who has been closely following the evolution of this profession since 1990 (CTV News, Reuters, The Canadian Press, Journal de Québec, and Huffington Post), I am pretty much in agreement with him. In Quebec, The Canadian Press (CP) is, for example, one of the rare media outlets to use AI to help with the translation of dispatches.

[photo] Journalism is not keeping pace with the evolution of new technology, according to Francesco Marconi, author of the book Newsmakers, Artificial Intelligence and the Future of Journalism..

AI does not replace journalists

Artificial intelligence is not there to replace journalists or eliminate jobs. Marconi believes that 8-12% of reporters' current tasks will be taken over by machines, which will in fact reorient the work of editors and journalists towards value-added content: long-form journalism, feature interviews, analysis, data-driven journalism, investigative journalism.

At the moment, AI robots perform basic tasks like writing texts two to six paragraphs in length on sports scores, quarterly company numbers, election and Olympic results. The outcome is convincing but also shows the limits of AI. We will always need a journalist to enhance a four-paragraph article on Bombardier's financial results, for example.

The analysis of large databases by AI robots also allows journalists from the Bloomberg news agency to receive an alert as soon as a trend or anomaly emerges from big data.

For Marconi, AI can also save reporters a lot of time by transcribing audio and video interviews. The same is true for major reports on pollution or violence, which rely on vast databases. The machines can analyze complex data in no time at all.

Afterwards, the journalist does his or her essential work of fact-checking, analyzing, contextualizing and gathering information. AI can hardly replace this. In this sense, the human being must remain central to the entire journalistic process.

A broken business model

Marconi is quite right when he explains that the media must develop a paid subscription model, get closer to their community with even more relevant content, develop new products (newsletters, events, podcasts, videos) and new content that AI can facilitate: personalized news, recommendations for readers, for example.

Some examples in Marconi's book are quite simple, like automated lists or articles about new local restaurants or businesses, which are very popular in the United States.

In this sense, AI is part of a new business model based on breaking down media silos. There needs to be a symbiosis in the sense of a "close collaboration" between the editorial staff and other media teams such as engineers, computer scientists, statisticians, sales or marketing staff.

In a newsroom, more than ever before, databases must be used to find stories that are relevant to readers, listeners, viewers and Internet users.

And there are already various AI tools available to detect trends or hot topics on the Internet and social media such as Dataminr, Newswhip, Parsely, Crowdtangle or Croma. These tools can also help newsrooms to better distribute content.

Beware of bias

Of course, the size of the newsrooms must be taken into account in all this analysis. A small weekly or a hyper-local media organization may not have the means to act quickly. But for the others, it's important to start taking action right away. Journalists need to be better trained internally; work with start-ups and universities to get the best out of the situation.

Let's take the current example of Covid-19. This is an opportunity to analyze public health data to make connections, analyze and dig into the data neighborhood by neighborhood and street by street. AI can help with that. But it takes well-trained data reporters to do this work.

One of the dangers of AI, on the other hand, is algorithm bias. Because algorithms are designed by humans, there are necessarily always biases that can alter data analysis and lead to serious consequences, according to Marconi. And human verification of content before publication will always remain a safeguard against errors.

An investigation by the U.S. media outlet ProPublica, which is funded through donations, showed in 2016 that algorithms used by the state to adjudicate parole cases resulted in a clear bias in favor of white inmates at the expense of black inmates. When you think about it, this use of an algorithm has led to glaring injustices.

AI has also developed systems for detecting fake videos (Deepfakes) and fake news, which are of course supported by experienced journalists from Reuters and AFP, for example. This is good news.

In this sense, the transformation of newsrooms is only just beginning and Marconi's essay is a must-read for identifying survival scenarios as media organizations and journalists. Because that's what it's all about. We need to better equip our newsrooms and completely rethink the workflow to achieve better collaboration and better content that will attract new paying subscribers.

Text by Patrick White, professor of journalism at the University of Quebec in Montreal (UQAM), and independent journalist.