Could Robots Write Our News for Us?

Concerns about technology replacing human jobs have existed since the 18th century. Yet last month’s announcement that the Press Association had received a Google grant for automated content production alarmed many journalists and copywriters. Could machines generate our news in future, and how exactly would this work?

Rise of the Robots

Machine learning algorithms are constantly refining their output. The principle behind automated content is known as artificial superintelligence, epitomised by voice-controlled chatbots like Alexa and Siri. Every interaction advances the software, delivering more accurate responses.

The same principles could apply to robotic news production. Humans edit initial stories, identifying grammatical errors and flagging up problems. Our involvement gradually diminishes with each correction, as the algorithm’s knowledge and consistency increase.

Press the Right Buttons

Automated content has been metaphorically making the news for decades. In 1993, an erotic thriller called Just This Once was co-written by a Macintosh computer. In 2010, Narrative Science began developing an Advanced Natural Language Generation platform that’s since produced content for leading blue-chip companies. Rival platform Automated Insights produced 1.5 billion ‘narratives’ in 2016 – the same year algorithmic text made it through the opening round of Japan’s prestigious Hoshi Shinichi Literary Awards.

From a publisher’s perspective, automated content offers many benefits: it’s far cheaper than employing humans for a start. Also, AI won’t deviate from specified grammatical rules, which ensures consistency. Computers can instantly respond to breaking news. And machines don’t add bias or skew stories the way humans do.

Superintelligent – but not Sentient

So are journalists an endangered species? Not necessarily. Machines are great at interpreting statistical data, but they’d struggle to create opinion pieces or add context and humour to their reporting. Robots can’t compassionately interpret the consequences of a military assault, or judge when to press evasive interviewees on unsatisfactory responses.

Tomorrow’s algorithms will probably generate framework text for copywriters to refine and expand upon. Nonetheless, don’t be surprised if news updates and sports reports begin emanating from algorithms rather than journalists in future…