Every newsroom has someone whose unremittingly joy-free outlook on life has made them a soundalike for Marvin the Paranoid Android, writes LAURA SLATTERY

CAN ROBOTS replace journalists? If a robot had written this article, it would probably answer "no" and move swiftly on. But the question, posed by Wired.co.ukassociate editor Olivia Solon at Tuesday's Media Future conference, is a more awkwardly reasonable one than it might at first sound to, ahem, humanoid journalists.

Robots, or artificial intelligence to be less earthling about it, are already capable of generating sports and financial reports, albeit ones that make even highly formulaic flesh-and-blood print and newswire offerings look like daringly experimental, idiosyncratic literature.

Take, for example, the corporate earnings previews published on Forbes.comby Narrative Science, a Chicago-based technology company that makes "stories and insights" from "structured data sources" and does so "at a scale that is simply unreachable given human resources alone".

Its earnings preview for News Corp’s recent third-quarter results, just one of many similar reports crafted from its algorithms, is all terribly accurate, noting a rise in analysts’ “buy” ratings, climbing revenues and advancing share price. It’s also personality-free, which is to say that it mentions Rupert Murdoch zero times. If you didn’t know it was generated artificially from a fixed number of financial performance metrics and sentence variables, you would come away from the report thinking News Corp was itself run by slick machines.

Indeed, the Narrative Science treatment of News Corp highlights a dimension of the company that sometimes gets lost in the mix of human-generated coverage - that so far, it has remained financially untouched by the phone-hacking scandal, give or take the $258 million in “fallout” costs reported by Bloomberg. To human eyes, there is something missing in Narrative Science’s preview. Its audience, however, is an equity investment community that is itself fond of using automated programs – arguably, its end-users are other computers.

Machines are good at the “heavy lifting”, says Solon. They can trawl through vast amounts of data to generate a loose form of narrative. Anything that’s up or down, buy or sell, or “fairly clear cut” has the potential to be automated. Or as Solon puts it: “Most of the articles that we let robots take over are fairly dreary for journalists.”

It’s the kind of glamour-light information processing that represents a commercially important segment of journalism, but hardly represents the occupation as a whole.

“I’m not sure a robot would be able to come up with headlines for the Sun,” says Solon, citing as her tabloid favourite that paper’s 1998 “Zip me up before you go go” splash after George Michael was arrested following an incident in a public toilet.

Although concepts such as right and wrong, happiness and sadness, personality and humour may be much trickier for machines, will they always prove so impossible? Tabloid puns are hardly so nuanced as to be beyond a robot forever. In any case, the importance of a news organisation’s ability to play on Wham! song titles may have already been diminished in the age of boring online headlines inspired by search engine optimisation.

Meanwhile, what Solon calls the “creepily realistic” Gemonoid F lady-robot, the invention of Japanese robot designer Hiroshi Ishiguro, boasts 65 facial expressions – that’s more than enough for all but the most banter-laden of television news bulletins.

Face.com's new Klik app, the first camera app with software that automatically tags photos of your Facebook friends, points to a potential use of similar technology in the automation of entire genres of reportage. Imagine the joys of combining real-time facial recognition with what has long been regarded as the technological Holy Grail for journalists: software that can reliably translate hours of audio recordings into words on the screen.

These are all just convenient tools, “dangerous in the wrong hands”, as Solon notes. They might speed things up for time-pressed editors and reporters, assist with the formation of logical arguments and cost a few more old-school jobs along the way, but human judgment is still required at some point along the “content creator” chain. Even at Narrative Science, its algorithms use vocabulary originally compiled by journalists known as “meta-writers”.

Every newsroom has someone whose unremittingly joy-free outlook on life has made them a soundalike for Marvin the Paranoid Android, the depressed robot created by the human brain of the late Douglas Adams. That doesn’t mean the media is turning completely Cylon-like just yet. “At least, for now, I think we can probably be friends,” says Solon archly of the media’s machine servants. “But I think we need to sleep with one eye open.”

The last thing anyone in the industry wants is the droids in sector G joining the robot union, bleeping that their corporation is being led by too many carbon-based life forms of insufficient processing capacity, and then plotting to overthrow their human overlords.

If we’ve learned anything at all from science fiction, it’s that the ultimate robot ambition is world domination. The scary question is not whether robots could ultimately replace journalists – it’s why would they even want to?