Think AI is bad for journalism? This story might change your mind: Letter from the Editor

Much of the reaction I receive from readers when I write about our newsroom’s experiments with artificial intelligence is negative, expressing fears that we’re handing over decisions and judgments to technology, removing the human touch.

That’s not what we’re doing. We see AI as a tool just like we see Microsoft Excel as a tool. We can use it to do our job better and more efficiently. It’s not about replacing the humans. It’s about helping them produce more content of value to you.

But, this might be a situation where it’s better to show, not tell.

Last weekend, we published a massively long story listing the purpose and status of all of Donald Trump’s executive orders since he became president in January. It was a big hit with readers, who shared it widely with their friends and family. It was so long we had to spread it over 9 pages in The Plain Dealer.

We could not have done it without artificial intelligence.

The project came about at the request of readers. Maybe a month after Trump took office, I started receiving emails and text messages asking us to publish the full, ever-growing executive order list. I received those requests from supporters and opponents of Trump. Supporters wanted to see the scope of changes as evidence Trump was making good through on campaign promises. Opponents wanted to see them as evidence he is destroying foundations of our republic.

As the number of orders grew, the job of compiling them became daunting. If I assigned a reporter to do it, the job would have taken weeks or more. But, as I continue to exhort our team to experiment with AI, about three weeks ago I tossed out the idea of using AI to help.

Leila Atassi leapt at the chance. Leila, the editor of our public interest and advocacy team as well as a columnist and regular on the Today in Ohio podcast discussion, is as big an advocate for AI in the newsroom as I am. She understands it will revolutionize newsgathering in ways we can’t yet comprehend, and because she has a lot of years left in her career, wants to be on top of the trend.

She collected the full text of all the executive orders – a massive amount of copy -- and ran them through an AI engine to summarize them. This was the most time-consuming part of the job, all that copying and pasting.

The result was too long for our purposes. The summaries were five and six sentences long. To make our piece accessible, we thought the summaries would have to be single sentences -- comprehensive while concise.

Back to the AI engine Leila went, and back she came with the summaries we needed. Leila was able to quickly doublecheck them for accuracy, but the amount of time saved by starting with the AI summaries – as opposed to digesting the orders and writing them ourselves – was enormous.

With the summaries in hand, we thought the reader would be better served if we could show the status of each one. To do that one by one would have taken forever, but asking the AI engine to do it was short work. The AI provided links to news sources for checking the status, and fairly quickly, Leila had a complete list of the orders, their summaries and, for each, their status.

Now, consider that this kind of work is not part of Leila’s job. She manages a team of reporters, edits most of our community columnists and, since the beginning of this year, writes her own weekly column, one that already is popular with our audience. Her working hours are spoken for with her normal duties. Four days a week, she appears on our podcast.

Because of AI, though, she managed to get this project together in the spare moments she has and on her own time at night. All told, she estimates she spent 8 hours on it.

The result was the most shared story in Sunday’s Plain Dealer and a stream of notes thanking us for the effort.

This is an example of the potential for AI. We can do more, do it faster and do it smartly, providing high-quality information you want.

When you see a note on a story saying that AI contributed, it does not mean AI did the work. Our reporters and editors are on top of every piece of content we publish, meaning that no matter how we put together the content, it remains us, the people and the institution you trust, putting it together.

In this case, we could give readers something they asked for, something we previously would not have had the resources to handle. With AI in our toolbox, we are a stronger newsroom.

I’m at cquinn@cleveland.com

Thanks for reading.