“We cannot have science in bits and pieces, applying it where we feel safe and ignoring it where we feel threatened”. This quote belongs to Carl Sagan, the acclaimed and Pulitzer prize-winning American author whose hundreds of papers are now archived in the Library of Congress in Washington D.C., the city we are visiting now as part of the WPI program. Sagan was a passionate scientist but also – and perhaps most importantly– a man who dedicated a big part of his life to questioning everything and advocating for people in general to do so.
Some days before our trip to D.C., South African journalist and WPI alumna Camilla Bath led a presentation for us at the University of St. Thomas. The subject was artificial intelligence (AI), a matter that has quickly become more present in everyday dialogues in newsrooms.
“Is AI going to replace journalists in the close future?” That’s a question I might have heard (and asked myself) way too many times in the past two years. Meanwhile, the world has witnessed how good AI has evolved to the point of becoming scary.
“No, AI is not replacing journalists,” Bath said in the classroom. But then she added a crucial point – we need to be careful and remain cautious of this technology.
Bath’s warning about AI reminded me of what Sagan wrote in his book “The Demon-haunted World” about science, which I quoted by the beginning of this article. If we focus on AI, we can find that this specific technology can be compelling and extensively useful for humanity, but at the same we cannot ignore its dark side. From deepfakes to information bias and unethical practices, AI has the power to cause global chaos, where anything could be framed as real and individuals could feel uncertainty while trying to get informed.
During our first immersive WPI fellowship weeks, we confirmed that AI can help journalism be stronger when applied well. In many cases, it can be a powerful ally in our work. Sunil Rajaraman is the founder of Hamlet, an AI tool that helps dissect information from official public meetings. Journalists who report on local policies spend hundreds of hours of their careers trying to identify what official public servants say during long – and sometimes uneventful– hearings. The entrepreneurs and creators of Hamlet AI saw this as an opportunity for profit, but also as a window to simplify research for the public interest.
And yet, AI can’t take all the credit in that work.
“You can’t rely on AI alone,” said Rajaraman when I asked him about the time invested by his team to refine Hamlet’s results. “So we actually have humans in the loop that kind of help us clean the data. We have distributed people across the U.S. that help us with data cleaning to get a 100% accuracy on results that we produce. So, it’s sort of like AI gives you 99% and for the rest you need people”
From a realistic perspective, no matter how refined AI could be or even how many gaps in that 1% disappear, the human work – and soul – cannot be replaced. Journalism is one of those things, and modern tools that help reporters do their job much more efficiently should be welcome with caution.
