Google testing Genesis AI for journalists: A brief history of tech in the newsroom

Google on 20 July showed a glimpse of a new AI tool, ‘Genesis’, to a limited set of executives from three top United States’ news publications — The New York Times, The Washington Post, and News Corp, owner of The Wall Street Journal. The tool, as reported by The New York Times, has the capability to process real-time events and data to create news articles.

According to the report, Google believes Genesis could act as a helpful personal assistant for journalists by automating certain tasks, allowing them to focus on other aspects of their work. The company views Genesis as responsible technology that could guide the publishing industry away from potential issues associated with generative AI. 

However, The New York Times also stated that some executives, present during Google's pitch, expressed concerns about the implications of this AI tool. They emphasised the importance of human effort in creating accurate and compelling news stories, raising worries about how Genesis could impact journalistic integrity. 

Taking this into account, let’s explore the most prominent instances of AI usage in newsrooms to date and assess how these trials have progressed.

Auto-content generation

Many media companies are now utilising Natural Language Generation (NLG) to transform structured data into written stories that closely resemble those written by humans. 

One such example is The Washington Post, which developed Heliograf in 2016, an in-house program that automatically generates short reports for their live blog. They initially used it during the Rio Olympics to share information like medal event results with services like Alexa. 

Over the course of a year, The Washington Post used Heliograf to publish 850 stories, expanding its use to cover subjects such as congressional races and high-school football games. 

Bloomberg and the Associated Press are also adopting NLG technology. They utilise Wordsmith by Automated Insights, a tool capable of mining data and producing reports on topics like sports results and company earnings. 

Use cases in fact-checking

Journalists are embracing AI technology to combat the spread of misinformation on social media and enhance their fact-checking processes. Among these tools is Chequeabot, a program developed in Argentina, which efficiently identifies fact-checkable claims in news articles and promptly sends them to newsrooms for verification, and automated fact-checking application. Claimbuster and Factmala are also being utilised.  

In past instances, AI fact-checking tools have been successfully deployed by major news organisations. Back in 2016, the Associated Press used Fact Check Tools to verify the accuracy of Donald Trump's claims about his tax returns. Similarly, in 2018, The Washington Post relied on AI fact-checking tools to analyse images and videos circulating on social media during the Parkland school shooting. 

The adoption of AI-driven fact-checking solutions comes as a response to the decline of trust in news caused by the prevalence of false information on social media platforms. 

Detecting search trends

Bloomberg and Financial Times are using AI to detect trends in news articles. They analyse the frequency of keywords and phrases to do this. Bloomberg's AI system can quickly identify trends in real time, helping them highlight the most important stories for their users. 

Moreover, Bloomberg utilises AI to determine the sentiment in news articles. The AI system analyses the tone of the writing to identify whether an article is positive, negative, or neutral. 

In addition to trend detection and sentiment analysis, Bloomberg makes use of AI tools for spell check and language translation.

Questions of transparency in use of AI in newsrooms

Several outlets have embraced generative AI or already integrated it into their content. CNET, for instance, faced criticism when it announced its use of AI to write news articles without initially disclosing it. This lack of transparency led some users to mistake the AI-generated articles for human-written ones, potentially impacting the trust in the accuracy and reliability of the information. In January, CNET decided to pause its controversial article-writing bot. 

Similarly, BuzzFeed also disclosed its utilisation of OpenAI’s ChatGPT to power its popular personality quizzes, while the New York Times employed ChatGPT to create a Valentine's Day message generator using a mix of prompts. 

According to a report from the Reuters Institute for the Study of Journalism, newsrooms have been adopting AI-powered tools to automate and enhance their processes. For instance, some digital portals use AI to display sports scores, market indicators, and other information. This trend is noticeable not only among global news agencies like Reuters, AFP, and AP but also among smaller outlets. 

AI in broadcast journalism

On July 20, Power TV, a well-known Kannada channel, introduced South India's first AI news anchor named “Soundarya”. Soundarya made her debut on a news show as an AI news presenter. 

Similarly, on July 13, Odisha TV, an Odia-based news station, also unveiled their AI news anchor, “Lisa”. In a video shared on Twitter by OTV, Lisa confidently introduced herself and expressed her excitement for this historic occasion. The news station revealed that Lisa would soon be hosting news updates, showcasing her capabilities as an AI news anchor. One of Lisa’s remarkable abilities is her proficiency in multiple languages, including Odia, English, and more.