The Associated Press news service announced that they plan to use computer generated stories. These articles will be generated using technology from a company called Automated Insights and will be used only for stories about corporate earnings reports. Their software uses complicated algorithms which take data and reorganizes it into prose. The company is going to get data from a financial news service called Zacks, and what they’re going to do is use it to turn out about 4.400 earnings reports stories. They currently do about 300 with their human staff.
Associated Press says that this is not about replacing human reporters with robots, this is more about radically increasing the number of stories these reporters can produce. They’re also saying they are not going to have job cuts or lessen the amount of business stories. In fact, this is going to free them up and allow them to do more enterprise stuff, more investigation, and do the kind of stories they couldn’t do when they were churning out all these earnings reports. This doesn’t actually sound like such a bad idea, provided the data is sound to begin with. It gives more information to more people because there’s more stories. These are the kind of stories that take journalists away from the more important ones. So many Associated Press journalists will most certainly be happy that robots will do it for them. Earnings reports stories don’t usually need an editorial feel anyway so there shouldn’t be much of a difference.
However, this company that they’re relying on, Automated Insights has a kind of radical vision for the future of journalism. They’re basically saying that right now there’s one story that’s going to a large number of people, but in the future they want large numbers of stories going to each individual person. In other words, each person will have their own version of every story. Maybe we’re exagerating a bit but still, it sounds like they’re on a mission to remake journalism and make it more automated and robotic. But as long as the articles are marked clearly there shouldn’t be any problems for the readers. There is already plenty of data that’s automated and the number will probably grow in the future. When you look at something like baseball scores for example, those are fairly easy to construct from data.
I imagine the Associated Press will be increasing the uses of this type of automation software for the creation of prose. We’ll probably see it creeping into other types of journalism, and we’ll likely see some mishaps along the way and some confused readers at some point. But if this enables a lot more information to be delivered to the public in an easily digestible way I think that’s a good thing at the end of the day.