On Friday, May 30th, I’ll be speaking at the Tow Center for Digital Journalism at Columbia University at the Quantifying Journalism: Metrics, Data and Computation Conference. I’ll be talking about the Robot Reporter trend, based on the four years of work I’ve done creating over 100 million automated stories via Automated Insights.
100 million stories, each one unique and professionally written, covering everything from news to finance to sports to marketing and beyond. In fact, by the time 2014 is over, Ai will have produced over 1 billion automated articles. And what I plan to tell the gathered Columbia students and assorted journalism professionals at said conference is that we’re not going to put any journalists out of work.
Well, any decent journalists, anyway.
I feel pretty good about that bold statement, basically because it’s one I’ve backed up continually over the last four years, although the need to defend our intentions, so to speak, has lessened recently. People are starting to get it now, the fact that automated content works best in situations where live human journalists either can’t produce the content, as in the case of the millions of fantasy football recaps we produce every Tuesday morning, or don’t want to, as in the compiling of mountains of big data into an easily digestible narrative.
When you consider that, automated content is actually another tool for the hardworking journalist, not competition.
If anything, those who consider themselves data visualists — the Excel wranglers and infographic ninjas and Powerpoint enthusiasts of the world — those are the people who should be worried about automated content.
Oh, and you listicle folks, we’re definitely coming for you. With prejudice.
In those early days, when whether or not we would be replacing journalists came up in every single interview and press piece we did, we always sort of laughed it off, but we never actually denied it, because we knew that angle made for a very good story.
But also because we couldn’t deny it.
The truth is, automated content can already deliver news. In fact, in some cases automated content is seen as more trustworthy than human-generated content.
But can it do the work of a journalist? Can it chase down leads and tap existing contacts and provide expert analysis and go by instinct and artfully use words to elicit emotion.
No. Not now, anyway. And if and when it does, the role of the journalist will evolve into something that can’t be automated, because that’s how technology works and that’s how humans work.
The fact of the matter is the paths don’t actually cross.
Automated Insights started life as sports-focused StatSheet.com. When we first started hiring, we looked at former journalists for our content designers, and since automated content was a new science and since the number of sports journalists looking to leave sports journalism was plentiful, we went that route for our first few candidates.
It was a spectacular failure.
Human journalists didn’t work because human journalists couldn’t stop looking for the qualitative story in the data, and when they couldn’t find it, they usually just gave up. It was like making them work their way through a dark room with blindfolds and earplugs.
But those stories are there. You just have to know how to get them.
The story that you get with automated content comes with the data, and while it’s a very good story and in some cases can even replicate a human-driven story, how you get there is a completely different path.
In other words, we can tell you everything you need to know about the player’s slump, but the set of data that colors the context of the slump can’t source the adversity that might be causing it. You have to go somewhere else for that. In our case, that’s probably another set of data.
And that’s the root of what a journalist does, using the hard facts to look for the deeper story. Automated content is there to make it easier to gather supporting data for the reporting of those hard facts.
Any journalist worthy of the title should get this, and welcome automated content as the robot that can do 90% of the stuff they can’t and don’t want to.
Which is the point of automation and technology anyway.