Trading floor of the New York Stock Exchange, 1963
Since writing a piece back in March about the possible intersections of photography and big data practices, the topic of ever smarter algorithms and their potential impact on our lives has remained constantly in my mind. It’s easy to see this topic as very abstract one without much practical grounding, but almost everywhere I look I see clues about how important this is going to become in the coming years for all of us, almost regardless of our occupations. Lately I’ve been thinking in particular about how the automation of data and image production and interpretation might come to further affect the field of journalism, for better and for worse.
I say ‘further affect’ because despite the commonly held belief that creative industries like journalism are immune from what Keynes termed ‘technological unemployment’ there is plenty of evidence that suggests otherwise. There is no doubt that algorithms already are making substantial inroads into journalism. Unsurprisingly this has mostly occurred in the most statistic and data heavy areas of journalism, and in those where writing is arguably least creative. The best known example is probably Quakebot, developed in 2012 to write generic stories based on recent earthquakes for the LA Times, using data gleaned directly from the US Geological Survey.
These quake articles are hardly exciting reads, but algorithms producing prosaic articles for public consumption are clearly a stepping stones to algorithms producing much more complex, engaging writing. Increasingly business and sports articles are being generated by algorithms, which again show no great flair for journalistic creativity but do show a step up in sophistication from generic reports about recent quakes. Meanwhile companies like Automated Insights are developing algorithms which have a much more sophisticated grasp of narrative and how to turn piles of raw data into a report a human might actually want to take some time over reading.
The increasing sophistication of algorithms is giving rise to some fascinating debates about the extent to which an algorithm can be said to display creativity, or whether this is a category which is exclusive to the human brain. Some of these debates are demonstrating just how subjective the idea of ‘creativity’ can be, and how dependent it can be on perception. People played an algorithmically composed symphony have sometimes been held rapt or reduced to tears by the genius of it, but some of the same people then react dismissively to that same piece on discovering it wasn’t composed by a human but by a string of code. It’s all sounds a bit sci-fi but this opens up interesting question about how prejudices might shape our relationship with algorithmically generated material.
Moving back to journalism, what many people seem to be asking is are these journalistic algorithms a blip, or are they the future? I can see plenty of arguments for the latter answer and not many for the former. As in other fields that have been enormously disrupted by the advent of algorithms, their increasing suitability to journalism seems inevitable if only because of the economics of the industry. Falling revenues, shrinking budgets and margins and corners cut or shaved where-ever possible are the reality of much modern journalism. In such an environment why would you hire a human journalism or an editor if an algorithm could do the same job?
That’s particularly true when talking about time critical assignments, where an algorithm might be able to produce a finished story in the time it takes a journalist just to sit down. Quakebot for example is reportedly able to have an article finished about a tremor almost at the time that the earthquake itself is ending , and similarly provided a steady flow of real time stats some sports algorithms are able to file a finished story on a game as the final whistle blows. In an industry where employers place more and more emphasis on efficiency, and audiences expect news to be ever more timely, the shift towards greater use of algorithms to write articles seems inevitable.
It’s not all doom though, and I think it’s possible to detect great benefits resulting from the advent of algorithms, even for journalists (at least those that survive). With growth of data journalism and the common reliance on crowd-sourcing in order to filter and interpret the massive data sets that news outlets are increasingly turning to as part of their investigative works, an increase in automation is obviously appealing. Some even argue that algorithms would be an advantage because the lack the inherent biases of journalists, maybe a slightly unconvincing claim that understates the extent to which algorithms can inherit the biases of their programmers, and the extent to which certain biases are appealing to some news outlets (which in turn opens up the slightly scary idea that journalistic algorithms might one day be programmed to reflect the biases of their readership).
Any competition between human and algorithmic journalists would also pose another interesting question. For now algorithms are being constructed to match existing human modes and practices of journalism, but there is the possibility that as they become more widespread our notion of what journalism is might change accordingly. We might see a divergence of journalism, as human writers are increasingly corralled into the more creative corners of writing and algorithms take on the majority of the day to day, relatively uncreative reporting. Or there might be a blending of the two, with algorithms assisting the choices journalists make, for example guiding editorial priorities based on reader responses, or helping a writer to generate those perfect click-bait headline that even reputable news organisations seem to increasingly chase.
In short I think all the clues suggest a profound and fascinating change is coming to the industry. Journalists can either do what many brokers did and try to ignore what is looming, or they can recognise it now and start to think about these change can be made to work for, and alongside them. In the follow up this piece to be published next week I’m going to take a closer look at photojournalism, a practice which for now might think itself almost completely immune from many of these innovations, but which I think will find in the coming years that it is as prone to disruptive technological change as any other.