On March 17, 2014, an employee at the Los Angeles Times was assigned to write a breaking-news article on the shallow 4.7 earthquake that had just hit California.
The reporter got to work in seconds, churning through data compiled by the U.S. Geological Survey. They pulled out key numbers, phrases and facts, and put the information into a template news article. This article was then sent to the content management system at the Los Angeles Times, where it awaited review and publication by a supervisor.
The supervisor, Ken Schwencke, then clicked ‘publish’ on the reporter’s article – much to the collective relief of the Los Angeles Times. It had been three minutes since the earthquake occurred and the news corporation was the first to publish a story on the event.
But it wasn’t the earthquake that garnered the popularity of the article; rather, it was the reporter. This was because the reporter wasn’t a human. Instead, it was an automated piece of algorithm software – a robot called Quakebot – which was programmed by Schwencke for the Los Angeles Times.
Quakebot’s role for the corporation was to generate a short article whenever an alert came from the U.S. Geological Survey that an earthquake above a certain threshold had occurred.
It wasn’t programmed to be creative or insightful; that job was left to the human staff after more information became available. Rather, Quakebot’s sole aim was to get information out to the public as quickly and as accurately as possible.
And it delivered.
The article – Quakebot’s first since its installation – was hailed as a pivotal point in journalistic history as it was the first automated news content produced outside of the corporate environment. It brought attention to the many robots at the frontline of automated reporting and the effects these could have on the journalistic profession.
AUTOMATED JOURNALISM TO DATE
The initial automated journalism movement was sparked through the creation of a news writing software by the technology company, Automated Insights, in 2011. The software, named Wordsmith, was the first major automated program to produce articles through the analysing of data.
Since then, the multinational news network Associated Press (AP) has used Wordsmith to produce roughly 3,000 corporate earnings reports per quarter – more than ten times the number its human reporters currently produce.
Wordsmith has also been commissioned to write recaps for Yahoo Sports Fantasy (American) Football, send GreatCall caretakers automated recaps on elderly citizens and create sales recommendations for Allstate.
And these stories all add up.
In 2013, Wordsmith produced 300 million stories – more than all other media outlets in the world combined. Wordsmith then went on to generate one billion stories in 2014 and the number is expected to climb even higher this year.
With so many stories being written by automated robots, the white collar job of journalism is now vulnerable to the prospect of automation. But will this automation actually encompass the journalism industry? And what will this mean for the future of journalism?
THE BENEFITS OF AUTOMATED JOURNALISM
Although we’d prefer not to believe it, more and more news articles are being generated by robots running on software algorithms. At the moment, such software is limited to formulaic journalism, such as sports, financial and political reporting.
News stories outside of a formulaic pattern are a struggle to generate, so for the time being, journalists still have the upper hand.
However, the journalism profession is still vulnerable to automation, due to the advertising-based business model that traditionally supported it failing to generate revenue in an online world. Media corporations are therefore looking to cut costs – and what better way to do this than to replace humans with automated robots?
For starters, automated robots are cost efficient, and once installed, require minimal maintenance or upgrades. Wordsmith itself costs an average of $750 per month, and can be applied to several journalistic divisions. Running costs are then restricted to electricity and broadband bills, which are slight. The average automated robot runs on the same amount of electricity as a computer.
Automated robots are also a reliable employee. They work nonstop when required and have zero chance of being unwell, tired, biased, opinionated, irritable or distracted.
And, perhaps most importantly, automated robots are fast.
“Where it might take a human several hours to write a report, Automated Insights can create narratives instantly,” says Asher Lipsitz from Wordsmith’s parent company, Automated Insights.
This is of extreme importance in the digital age, where speed has become the crucible of our jobs. Automated software can produce a full story in the style code of their given organization in milliseconds. This final product is then published automatically to multiple pre-set websites, where it can be immediately accessed by the public.
From here, the subtlest benefit of automation comes into play.
Studies have shown that ordinary citizens cannot differentiate between a publication authored by a human and a publication written by an automated robot. New York Times’ Human or Computer quiz proves this online, by getting its news readers to try and choose the author’s genus.
New York Times’ Human or Computer quiz tests your ability to spot a robot.
Although amusing now, it could hold dire implications for the journalism industry in the future. If robot-authored articles are indistinguishable from those that are human-authored, then will the public notice if journalists are being replaced?
SOFTWARE EXPANSION: PERSONALITY
Robot-generated stories are also advancing beyond the typical “fill-in-the-blank” algorithms.
“The amount of quantifiable data collecting is expanding exponentially,” says Mr Lipsitz.
“As we have more data on how the world sees, smells, looks and works, [software’s such as] Wordsmith will have increased capabilities and will write more complex narratives.”
The more advanced software’s are now using perspective, tone, humour and voice to tailor a story to suit specific audiences of its employer.
For example, an automated robot reporting on an NRL match between the Broncos and Cowboys is capable of producing two versions of a story, using the same data: one written for a Broncos fan; and one written for a Cowboys fan.
Additional aspects can be added from here, such as bravo, depending on who won, and humour, depending on events.
“With the world collecting more and more data, it is only logical for automated software to be able to use that data to create a story, or even facts that it would take humans a long time to uncover,” says Stuart Miles, tech-journalist and CEO of the gadget news website, Pocket-lint.
“The hard part is teaching the software to understand the data – even humans struggle with that a lot of the time.”
BRAIN BEATS BIONIC
But these robots, as sophisticated as they may be or become, could never attain the same level of individuality as the human mind.
Automated journalism is based on the idea that brains are essentially computers and therefore can be “substituted” for something more efficient. However, the computer metaphor ignores – arguably – the most species-defining characteristic of human beings: that we can intentionally “malfunction”.
As journalists, we know what we are supposed to do within our profession, but we have to option to (and often do) consciously disobey this “programming”.
Examples of this happen every day.
Something that may be legally or morally questionable when published – huge no-no’s in the journalistic world – may be worth the violation of journalistic ethics if it is in favour of the public’s interest.
Automated robots, on the other hand, are predictable by nature and will always fulfil what they are programmed to do. Even if a robot is designed to make irrational, illogical decisions, it still does so out of design rather than of its own omission.
So, no matter how advanced they may become, automated robots will never be able to surpass the individuality of the human brain – and therefore, the human journalist.
This isn’t the only factor that makes our professional future impervious. Another is the inability for robots to conduct in-depth, investigative reporting.
“Automated software can go through large dumps of data to find patterns and anomalies much quicker than people. But it can’t develop sources, it can’t reach out to whistle-blowers and it can’t develop trusted relationships with people in powerful positions – or positions of power,” says Amy Remeikis, Fairfax Digital’s Queensland political editor.
“Journalists still receive anonymous packages containing information, phone calls from blocked numbers asking for help, emails from dummy addresses telling them to look in a certain direction.
“While software can be programmed to go through government open data websites and search, it can’t put it in context as to why the anomalies or information impact democracy or people’s lives.
“While we can program software and machines to do a lot of our jobs, we are still a very, very long way away from replacing the human mind – and the gut feelings which often lead journalists down paths they didn’t know existed – and the tenacity to keep following that path in the face of denials.
“That’s where we get some of our best and most important stories – such as the reports which led to the Fitzgerald Inquiry. Software is a long way away from replicating that.”
IMPACT ON NEWS REPORTING
Despite the inability to replace human journalists, automated journalism is going to become more common – especially with the expansion of the digital age, where the economics and demand are too compelling to do otherwise.
So the question becomes not if it will happen, but what effect will automated journalism have on our profession when more of our news is being written by automated robots?
On one hand, it’s possible that by freeing up human intellect from the same repetitive stories, reporters will be able to focus on more important journalistic activities.
“I think the boring, repetitive tasks will be removed, forcing journalists to be a lot more creative in their approach,” says Mr Miles.
“We are already seeing automation for reporting on stocks or the weather, for example, and that will only continue to grow. It’s the boring side of the work. One that needs to be factually correct, but not one that needs to have exciting or interesting prose.”
Mr Lipsitz agrees, stating automated robot’s will “allow [journalists] to focus on providing the most accurate and important information for readers”.
“Articles on factual information like earnings reports and box scores, that journalists used to write, are often done through Wordsmith now,” says Mr Lipsitz.
“So far, we’ve heard from journalists that they enjoy using Wordsmith because it opens them up to other, more interesting work to produce.”
On the other hand, the introduction of automated journalism might encourage a higher-quality form of journalism.
Robots might not be our “competitors”, but our full-time research assistants, who can prepare data-base stories that we can supplement with our own thoughts and opinions. This trans-human mergence has the potential to produce articles that may be better than the efforts of either man or machine alone.
“Journalists often use Wordsmith as a tool more than a replacement,” continues Mr Lipsitz.
“They can add or take out information from the narratives as they see fit, while focusing their time on their important watchdog function for the public.”
Geoff Crockett, General Manager for Centro’s Production Unit at APN News and Media, state’s automated software’s are a “great way of taking the limited resources we have in our newsrooms and applying them to better serve the reader”.
“You get the right format, you get the right size, everybody gets the same sort of treatment, everyone knows what’s expected, and for the reader, you’d get consistency,” Mr Crockett says.
“I can tell you that that would be a benefit.”
OUR AUTOMATED FUTURE
In the future, automated journalism is expected to begin a new dawn of journalism when it merges with other technological advances, such as individual tracking, browsing habits and mobile connections.
This combination of high-speed news and individual information is predicted to create a journalism profession in which articles are specifically drafted for individuals of society.
“Wordsmith is able to craft narratives that can target audiences much smaller than any traditional news organizations can,” says Lipsitz.
“Instead of writing a single article that might apply to a few thousand people, Wordsmith writes thousands of articles that apply to each individual person.”
News outlets will know who you are – from your interests to the model of your car – and have the “manpower” (i.e. automated robots) to redesign a news story based off your profile. This means that although there will only be one reader per article, more people will potentially engage in news items, because they will appeal to each individuals’ needs.
“We know that stories which resonate with readers have a very high WIIFM (what’s in it for me) component, which means that if a reader feels they are connecting with any particular story, they’re more inclined to wade their way through it, engage with it, respond to it, or share it,” says Simon Holt, Brisbane Times’ Editor in Chief.
“If we are then able to drill news down to the individual interests of each reader, we know that we’ll have a much higher engagement level.
“If we are to continue with an advertising-based business model, that’s good news for the industry. If our readers are happy, there’s every chance we’ll remain successful. If we aren’t providing what they can get elsewhere, our business will fail.
“It’s a simple equation, and customized publishing is a step towards a positive outcome.”
However, Tim Dunlop, author of The New Front Page: New Media and the Rise of the Audience, warns the predictive metrics that fuel technologies such as individual tracking and browsing habits, could bring their own set of issues in the future.
“Metrics often tell you about the past and then the publishers use [that information] to make predictions about the future,” he says.
“But just because people were interested in ‘x’ at this point in time, doesn’t mean people will be interested in ‘x’ on an ongoing basis. And the risk is that it closes down other areas of interest for people.”
These are issues we will have to address in order to prevent a journalistic profession based on limited selection. Until then, automated robots will continue to assist journalists and create a new era of mass-produced media.
It will be interesting to see where the future takes us.