How I learned to stop worrying and love AI
AI isn't replacing journalists. In fact, it can significantly help them.
After OpenAI shook the world in November of 2022 with its public introduction of ChatGPT, discussions of artificial intelligence’s intersection with the media were inescapable. Indeed, there is much to ponder. Particularly in an era when powerful people are playing fast and loose with a little thing I like to call “reality,” there is justifiable concern for AI’s potential to further destabilize the world through the creation of deep fakes and inaccurate information.
Those concerns are legitimate and I share them. Pretty much any large-scale discussion about AI when I was at The Washington Post ultimately gravitated to ethics, our efforts to authenticate media and fight misinformation. Those issues aren’t going away anytime soon because neither is AI.
But while organizations and our governing bodies wrestle with the potential perils of this new reality (thoughts and prayers to those tasked with briefing our geriatric congress on these matters), I personally think it makes sense to seize upon the upside AI offers. The more people can utilize programs like ChatGPT as useful tools, the more efficient and more successful they’re likely to be at their job.
Efficiency is a good place to focus because that’s one of AI’s primary merits. It’s a time saver both for creators and consumers.
I am by no means an AI expert —though I did use ChatGPT to author updates for my family’s fantasy football league using the tone and verbiage of Perd Hapley — and I’m also no engineer, so implementing these ideas requires additional skills. Still, by focusing on certain aspects of AI’s offerings, we can see some smart places media companies can apply it without eroding any ethical standards.
It’s a powerful aggregator, so let it aggregate
The Washington Post has a newsletter called “The Seven” in which the seven most-essential news items of the day are synopsized for easy, rapid digestion each morning. But there are far more than seven interesting stories in The Post everyday. (I’m biased, but I think you’ll agree with me.) AI has the ability to scan a news organization’s entire output each day and spit out a newsletter that collects headlines and a description for each story. It could take “The Seven” and make it “The Seven Hundred.” Would that be a great user experience? Maybe not. But if you integrate some other aspects of AI, you can start to see how it could shape up nicely.
AI can spot audience trends and react to them
Believe it or not, AI didn’t just come along in 2022. It’s been a key component of internet search and recommendation engines for decades. Think about the Amazon recommendation algorithm … that’s AI. It takes your browser history and customer data and pulls up products it believes you’re most likely to purchase.
Now, using our newsletter example above, AI can scrape a reader’s interactions with a site to discern what types of stories they tend to read. So from among those 700 stories, the top results would be topics with which they tend to engage. It’s kind of like creating a Spotify playlist.
This helps make sure readers are seeing content they like and helps make clear the value of that content to the consumer. The customer is now more easily getting what they want from their subscription.
Modern newsrooms pump out tons of content every day. But there’s only one homepage and its real estate is finite. Add in the essential coverage on topics like the election, Israel and Gaza, Ukraine and Trump’s many, many legal battles and there’s not much room left for, say, a really good profile on UConn men’s basketball coach Dan Hurley. Even a big college basketball fan might miss it on the site. (Like I did.) Incorporating audience data can help curate a more desirable, dynamic site and/or newsletter experience and make sure readers don’t miss stories they’re likely to enjoy.
AI can be an investigative team’s dream
Zach Seward, recently appointed as the editorial director of AI initiates at the New York Times, gave a talk at SXSW last month about the incorporation of AI and journalism. After outlining some of the very public failures as cautionary takes, he delved into success stories. This included using an AI program to help scan thousands of dense financial documents for an investigative story by Quartz on the tax haven of Mauritius.
The program, Seward said, was able to spot patterns that no human would be able to recognize, helping them find additional documents to aid their investigation.
Among several other examples of using AI to aid investigative journalism, the Wall Street Journal used AI image searches of Google Streetview photos to help spot lead cabling in power lines for a story about the disturbing prevalence of the toxic substance around the country.
About those pitfalls though …
Here’s why I’m not worried (yet) about AI replacing journalists: AI remains messy. Ask ChatGPT who founded Launcher and here’s what you get:
So much for my legacy. While this might seem logical and accurate to an uninformed inquirer, the information is wrong. While Christian definitely covers space for The Post, Launcher was our video game coverage vertical (RIP). What’s more, this answer hasn’t even been consistent. Asked previously, it’s noted the site was founded by an array of different journalists, some of whom have never even worked at The Post.
While AI can aid in a number of ways, it requires oversight. Just ask CNET or Sports Illustrated what happens when no one minds the machines. But by exercising the appropriate safeguards of fact-checking and editing, something newsrooms do routinely anyway, the incorporation of AI can be a valuable tool in producing top-tier journalism.
Coming Wednesday: Kim Mulkey’s really terrible PR move
Hi, I’m Mike. I’m a former editor for The Washington Post and ESPN. In 2024 I founded and now operate Launcher, LLC, a digital media consultancy operating out of Arlington, Va. Want to work together? Reach out on LinkedIn.