Artificial intelligence is here. “There’s already a volume of literature proving that it does aid journalism work. It’s up to us to find a way to use it as a leverage for our reporting.”
On the second day of the Philippine Center for Investigative Journalism’s (PCIJ) Third National Conference on Investigative Journalism on May 1, PCIJ editor-at-large Karol Ilagan presented how a few newsrooms in Southeast Asia have been slowly adapting AI for specific tasks or integrating the technology into their workflows.
At the plenary session on AI: Opportunities and Challenges for Investigative Journalism, Ilagan, Rappler’s Gemma Mendoza, data scientist Christian Alis and data analyst Dominic Ligot discussed the technology’s growing influence in society and the need for media practitioners to ride its waves.
Veteran journalist John Nery, who moderated the panel, kicked off the session by illustrating the capabilities of generative AI. He introduced the speakers with words and photos conjured up by AI. Other panelists pulled up videos that were manipulated to make its subjects speak in another language and utter fabricated sentences.
“Ever since [generative] AI made its debut, it’s been one ethical problem after another,” said Ligot, founder of Data Ethics PH. He and other panelists, however, were quick to dispel fears of AI’s looming “threat” to journalists’ jobs.
They highlighted how journalists can optimize AI to boost the quality of their content and efficiency in the newsroom. They also offered tips on how to navigate the technology’s ethical landmines.
‘Perfect Sounding Board’
Alis, a professor at the Asian Institute of Management, said journalists can use AI in transcribing their interviews and sifting through voluminous documents. Newsrooms can also determine how to best market their content through AI, he added.
Tackling the job’s editorial aspect, Ligot explained that AI chatbots can be a “perfect sounding board” as these help writers determine weaknesses in their articles. The key, he said, is “prompt engineering” or the act of meticulously prompting AI chatbots.
“If you add more context, you give questions, you give examples, it gets smarter,” he said.
For Ligot, one of the best uses of AI is creating chatbots tailored to specific uses, and out of existing content. This is known as “retrieval augmented generation.” A Filipino scholar and journalist, Jaemark Tordecilla, has already developed an AI tool to find stories from Commission on Audit reports.
Ligot identified more aspects of journalism that can be innovated by AI: content planning and brainstorming and content creation, review and analysis.
Still, he cautioned that “just because multiple tasks in the value chain of journalism can already be automated, [doesn’t mean that] journalists can be automated out. The way these models work require someone to prompt them or instruct them.”
Transparency
AI’s knowledge is limited to data available online, panelists impressed on participants. For this reason, it may not be reliable for discovering new information especially for in-depth stories.
“When we define investigative reporting, this is a product of our original reporting… So technically, if you ask ChatGPT, what we’re looking for might not be there yet,” explained Ilagan, also a journalism professor at the University of the Philippines.
Still, AI can offer endless possibilities in the field.
“At least four of [the International Consortium of Investigative Journalists’] global investigations used AI. They used machine learning to detect patterns from records [and] computer vision as well as natural language,” said Ilagan, a seasoned investigative journalist herself.
Meanwhile, Mendoza shared how their newsroom created almost 50,000 candidate profiles during the 2022 national elections with the help of Open AI’s GPT 3.
It needed to be done in a way where you minimize the possibility of it coming up with weird stuff. We limited it with structured data we verified ourselves.
Gemma Mendoza, HEAD of Rappler.com’s Digital Services & Lead Researcher, Disinformation & PlatformS
“It needed to be done in a way where you minimize the possibility of it coming up with weird stuff. We limited it with structured data we verified ourselves,” she said as she discussed the importance of human oversight when using AI. Rappler articles are also now prefaced with AI-generated summary points.
The news executive added that in every step taken with AI, the newsroom strives to be transparent: “Every time something is produced with AI it needs to be clear and the public needs to know.” Rappler’s AI guidelines are available on their website.
Dangers
After discussing how AI opens up possibilities for the media, the panelists drew attention to its dangers.
“Like any person can be fooled, these models are quite naive about things,” Ligot said. He highlighted issues on copyrighting, data privacy and cheating attached to AI. The need for industry-initiated guidelines that could govern how this technology should be used were also emphasized during the session.
A charter on AI and journalism has already been published by the Reporters Without Borders, Ilagan shared.
The charter aims to set fundamental ethical principles to protect the integrity of news and information in the age of AI, when these technologies are poised to transform the media industry.
But another factor needs to be addressed. Participating journalists wondered how newsrooms averse to AI and are lagging behind in digitalization efforts can be engaged.
For Ilagan, the first step is to make colleagues realize that the technology is already present. “Obviously, there are challenges and opportunities. But, again, you have to practice with it so you can understand on your own that it’s simply a tool,” she added.
For Mendoza, it should begin with conversations in the newsroom.
“As long as it is clear and it is discussed between management and the staff, what is the policy of the newsroom when it comes to AI… you have the means to [move] forward,” she said.