Writing and AI or is a robot writing this blog?

In an interview almost 10 years ago for an article I was writing for a digital publishing magazine, I had a conversation with a very technically oriented individually that went roughly this way,

Him: (enthused and excited) We’re developing algorithms that will let us automatically create brochures, written reports, that will always have the right data and can be instantly updated.

Me: (pause)

Him: (no reaction)

Me: (breaking long pause) You realize you’re talking to a writer, eh? You’ve just told me that at some point in the future nobody will need writers.

Him: (pause) No. (then with more certainty) No. You don’t understand. We’re making things better for you. In the future, you won’t need to do the boring stuff.

It seems the future is now and in the hands of a company known as Automated Insights, You can find this at the base of one of the company’s news releases,

ABOUT AUTOMATED INSIGHTS, INC.

Automated Insights (Ai) transforms Big Data into written reports with the depth of analysis, personality and variability of a human writer. In 2014, Ai and its patented Wordsmith platform will produce over 1 billion personalized reports for clients like Yahoo!, The Associated Press, the NFL, and Edmunds.com. [emphasis mine] The Wordsmith platform uses artificial intelligence to dynamically spot patterns and trends in raw data and then describe those findings in plain English. Wordsmith authors insightful, personalized reports around individual user data at unprecedented scale and in real-time. Automated Insights also offers applications that run on its Wordsmith platform, including the recently launched Wordsmith for Marketing, which enables marketing agencies to automate reporting for clients. Learn more at http://automatedinsights.com.

In the wake of the June 30, 2014 deal with Associated Press, there has been a flurry of media interest especially from writers who seem to have largely concluded that the robots will do the boring stuff and free human writers to do creative, innovative work. A July 2, 2014 news item on FoxNews.com provides more details about the deal,

The Associated Press, the largest American-based news agency in the world, will now use story-writing software to produce U.S. corporate earnings stories.

In a recent blog post post AP Managing Editor Lou Ferarra explained that the software is capable of producing these stories, which are largely technical financial reports that range from 150 to 300 words, in “roughly the same time that it takes our reporters.” [emphasis mine]

AP staff members will initially edit the software-produced reports, but the agency hopes the process will soon be fully automated.

The Wordsmith software constructs narratives in plain English by using algorithms to analyze trends and patterns in a set of data and place them in an appropriate context depending on the nature of the story.

Representatives for the Associated Press have assured anyone who fears robots are making journalists obsolete that Wordsmith will not be taking the jobs of staffers. “We are going to use our brains and time in more enterprising ways during earnings season” Ferarra wrote, in the blog pos. “This is about using technology to free journalists to do more journalism and less data processing, not about eliminating jobs. [emphasis mine]

Russell Brandon’s July 11, 2014 article for The Verge provides more technical detail and context for this emerging field,

Last week, the Associated Press announced it would be automating its articles on quarterly earnings reports. Instead of 300 articles written by humans, the company’s new software will write 4,400 of them, each formatted for AP style, in mere seconds. It’s not the first time a company has tried out automatic writing: last year, a reporter at The LA Times wrote an automated earthquake-reporting program that combined prewritten sentences with automatic seismograph reports to report quakes just seconds after they happen. The natural language-generation company Narrative Science has been churning out automated sports reporting for years.

It appears that AP Managing Editor Lou Ferarra doesn’t know how long it takes to write 150 to 300 words (“roughly the same time that it takes our reporters”) or perhaps he or she wanted to ‘soften’ the news’s possible impact. Getting back to the technical aspects in Brandon’s article,

… So how do you make a robot that writes sentences?

In the case of AP style, a lot of the work has already been done. Every Associated Press article already comes with a clear, direct opening and a structure that spirals out from there. All the algorithm needs to do is code in the same reasoning a reporter might employ. Algorithms detect the most volatile or newsworthy shift in a given earnings report and slot that in as the lede. Circling outward, the program might sense that a certain topic has already been covered recently and decide it’s better to talk about something else. …

The staffers who keep the copy fresh are scribes and coders in equal measure. (Allen [Automated Insights CEO Robbie Allen] says he looks for “stats majors who worked on the school paper.”) They’re not writers in the traditional sense — most of the language work is done beforehand, long before the data is available — but each job requires close attention. For sports articles, the Automated Insights team does all its work during the off-season and then watches the articles write themselves from the sidelines, as soon as each game’s results are available. “I’m often quite surprised by the result,” says Joe Procopio, the company’s head of product engineering. “There might be four or five variables that determine what that lead sentence looks like.” …

A July 11, 2014 article by Catherine Taibi for Huffington Post offers a summary of the current ‘robot/writer’ situation (Automated Insights is not the only company offering this service) along with many links including one to this July 11, 2014 article by Kevin Roose for New York Magazine where he shares what appears to be a widely held opinion and which echoes my interviewee of 10 years ago (Note: A link has been removed),

By this point, we’re no longer surprised when machines replace human workers in auto factories or electronics-manufacturing plants. That’s the norm. But we hoity-toity journalists had long assumed that our jobs were safe from automation. (We’re knowledge workers, after all.) So when the AP announced its new automated workforce, you could hear the panic spread to old-line news desks across the nation. Unplug the printers, Bob! The robots are coming!

I’m not an alarmist, though. In fact, I welcome our new robot colleagues. Not only am I not scared of losing my job to a piece of software, I think the introduction of automated reporting is the best thing to happen to journalists in a long time.

For one thing, humans still have the talent edge. At the moment, the software created by Automated Insights is only capable of generating certain types of news stories — namely, short stories that use structured data as an input, and whose output follows a regular pattern. …

Robot-generated stories aren’t all fill-in-the-blank jobs; the more advanced algorithms use things like perspective, tone, and humor to tailor a story to its audience. …

But these robots, as sophisticated as they are, can’t approach the full creativity of a human writer. They can’t contextualize Emmy snubs like Matt Zoller Seitz, assail opponents of Obamacare like Jonathan Chait, or collect summer-camp sex stories like Maureen O’Connor. My colleagues’ jobs (and mine, knock wood) are too complex for today’s artificial intelligence to handle; they require human skills like picking up the phone, piecing together data points from multiple sources, and drawing original, evidence-based conclusions. [emphasis mine]

The stories that today’s robots can write are, frankly, the kinds of stories that humans hate writing anyway. … [emphasis mine]

Despite his blithe assurances, there is a little anxiety expressed in this piece “My colleagues’ jobs (and mine, knock wood) are too complex for today’s artificial intelligence … .”

I too am feeling a little uncertain. For example, there’s this April 29, 2014 posting by Adam Long on the Automated Insights blog and I can’t help wondering how much was actually written by Long and how much by the company’s robots. After all the company proudly proclaims the blog is powered by Wordsmith Marketing. For that matter, I’m not that sure about the FoxNews.com piece, which has no byline.

For anyone interested in still more links and information, Automated Insights offers a listing of their press coverage here. Although it’s a bit dated now, there is an exhaustive May 22, 2013 posting by Tony Hirst on the OUseful.info blog which, despite the title: ‘Notes on Narrative Science and Automated Insights’, provides additional context for the work being done to automate the writing process since 2009.

For the record, this blog is not written by a robot. As for getting rid of the boring stuff, I can’t help but remember that part of how one learns any craft is by doing the boring, repetitive work needed to build skills.

One final and unrelated note, Automated Insights has done a nice piece of marketing with its name which abbreviates to Ai. One can’t help but be reminded of AI, a term connoting the field of artificial intelligence.

Leave a Reply

Your email address will not be published. Required fields are marked *