Insider announced that it will start experimenting with AI to write articles. CNET published AI-written articles for approximately 1% of its content. PR agencies are exploring AI tools that claim to write press releases and personalize pitches. AI will undeniably bring disruption to the newsroom ecosystem.
As we are figuring out what this means for our industry and ourselves, here are a few questions that come to mind.
Is AI capable of writing articles?
There’s always skepticism around anything new that requires a large behavior change. When personal computers and printers first emerged, there was a period before everyone gave up on the typewriter. Habits, dogma, and nostalgia, are a tough thing to break. We tend to overvalue the past and undervalue the future.
It’s generally been a mistake to judge technology based on where it exists today. GPT4, which was released last month, is 10x more advanced than GPT3. Part of the reason many technology leaders have asked to pause on the release of GPT5 is that it will in turn be 10x more advanced than GPT4. AI is already far more advanced than the average person in its reading comprehension abilities. When trying to understand the long term implications of AI, we should assume it will be able to do it all eventually.
We’ll be better prepared for the future if we don’t focus on the errors of AI today because the rate of evolution for AI proves that errors in logic will be resolved.
Is AI a threat to the value of journalists and journalism?
If all you care about are the words themselves then AI is going to win. We will get to a point where AI can make decisions that humans do now. This is already happening with some publications automating summaries of public company earnings reports and posting them as articles. We’ll simply see more of this happening with a wider breadth of content across an increasing number of publishers.
One of the results of using AI to generate content maximized for attention is that there will be an abundance of articles created that drives a race to the bottom. It’s the same reason everything became cat videos in the early stages of social virality. Whether or not we want to admit it, we clicked and watched a lot of those videos. AI will itself help to resolve this in not only the content it creates but the content it gets in front of you. Imagine you could customize an algorithmic feed based on a quality index.
All that being said, there are areas where it matters how the words came to be. It’s not the algorithmic feed or system determining value. Who decided it was worth writing? How did they source the content? Was it an investigative piece? How many people were interviewed sourcing information for the article? If we really think about why earned media has value over paid, it’s the human decisions leading up to the article that make it so valuable. The prestige of news value from a respected third party. That is not going anywhere. In fact, it will become (in some ways) even more valuable because it will become even more scarce.
Should we allow AI into the newsroom?
Depending on your definition of AI, it’s already been there. The articles on most homepages and sites are algorithmically curated. Headlines are algorithmically tested to enhance clicks. Many writers use tools like Grammarly to edit articles. A number of sites like SeekingAlpha have been writing automated articles based on earnings reports for a while. Taking steps toward AI-driven newsrooms might be considered a relatively thin line to cross.