Will AI Replace Journalists or Force Them to Do Their Job Again?
Nearly two years ago, OpenAI released ChatGPT to the world and ushered in the age of artificial intelligence (AI). ChatGPT, a large language model (LLM), offered users a true taste of AI’s potential to program code, analyze information, edit documents, and even just converse. Governments, businesses, and the media alike quickly began to realize the emerging technology’s potential to upend entire industries sooner rather than later.
Numerous LLMs and other generative AI models have since reached public markets. It is easier than ever to have AI write, edit, format, and fact-check content. The implications for journalism are clear: AI might soon become a cheaper and superior option for news organizations compared to a stable of human writers.
Why would media companies pay a full-time team of writers to find, research, and produce articles when they could instead have a smaller team of editors and reporters plug the relevant facts into an LLM at a fraction of the time and cost? Of course, some human touch is required before final approval of articles — given AI’s tendency to hallucinate information — but it is nevertheless easier to downsize newsrooms than ever.
I can speak from personal experience, having lost a writing job to AI. The clients who were ordering newsletter content from our team simply couldn’t justify the cost when AI is now an option. Personally, it was a reality check about just how quickly AI is transforming our world.
Even before the AI revolution began, newsrooms were shrinking. Newspapers in particular have been hit hard the last couple of decades. Many journalists have become independent or found work at think tanks, public relations firms, and nonprofits. As corporate, mainstream news outlets decline and alternative media sources rise, journalists are scrambling to find their place in a shaken-up system.
Anyone from nearly anywhere in the world can break a story on social media if they come across the right information at the right time and place, putting journalists at a disadvantage compared to the past. Some journalists have turned to podcasting, Substacks, and other forms of commentary, seeking to monetize opinions instead of “hard” news coverage.
Other journalists have started training the very AI systems that threaten to take their jobs. Training LLMs, which I have done as a side gig, often requires the same skills as journalism: good writing, accurate researching, and keen analysis.
AI developers frequently ask trainers to assess the ability of an LLM to respond to a prompt, judging it based on several criteria. The criteria often include whether the LLM satisfies all explicit and implicit requirements of a prompt, whether the information is accurate, and whether the writing is appropriately styled. Journalists are sought-after AI trainers because they are well equipped to teach AI how to respond to a prompt more like a human being, helping to insert a touch of mankind in the models.
They essentially serve the function of an editor in a newsroom — final fact-checking and style changes. Journalists who help train AI models may be digging their own graves, but they still have an upper hand on the technology. Although LLMs can write better than many journalists, real journalists can do something AI can’t: report and investigate on the ground.
For now, AI models lack a physical body to interact with the world around them. AI-powered robots are not yet wandering the streets of our cities in search of a scoop. Only a human with flesh-and-blood legs can prowl for breaking and exclusive news. AI can be a great writer when it is given the necessary facts and context, but it is operating on a delay compared to humans on the ground.
As more people turn to podcasts for their news, journalists have the chance to act as correspondents for booming online shows. Many podcasts are starting to livestream during special events, such as party conventions and Election Day, which gives journalists a chance to report for them on the ground, much like they would for a cable news show.
For example: on Election Day, I had the pleasure of reporting live from the US Capitol for Human ReAction’s livestream. I was able to bring unique insights straight from DC — something most podcasts covering the election did not have. Viewers love to see news from where it is happening, as it is happening. The demand for live, on-the-ground coverage is there, and displaced journalists provide a clear supply.
Soon, all the biggest and best politics podcasts will have teams of correspondents and reporters around the country. Just like mainstream media networks try to find scoops and break stories, so too do online content creators seek to be the first to cover urgent matters. If they want to stay ahead of the curve, they need to turn to the traditional journalistic practice of on-the-ground coverage.
AI will undoubtedly transform journalism, but it is not poised to kill it. As some work gets replaced, other work opens up. That is how industry after industry has changed throughout the course of human history in the face of technological advancements. AI will not replace journalists, it will only force them to return to the roots of the job.
