104 Turing’s Test
The most extensive survey of automated journalism at the time of writing is Andreas Graefe’s Guide to Automated Journalism for the Tow
Center for Digital Journalism at the Colombia University School of Journalism. Graefe considers the various means by which the use of algorithms
to gather news from highly structured sources offers an ability to produce
copy more quickly and with fewer errors than human journalists, while
also noting how it fuels those same journalists’ anxieties about their own
futures. Some of the latter issues are returned to by Carlson, who sees in
the example of the company Narrative Science a competition to automate
most if not all news content against journalists’ definitions of their own labour: very simply, software can produce a much greater number of stories,
but only when the news is defined within a strictly limited compositional
form.
16
Carlson runs through a series of headlines from a variety of publications in 2014 to 2015, such as “The Robot Journalist: Heralding an
Apocalypse for the News Industry?” (Guardian) and “Can Robots Run
the News?” (Mashable)
17
– demonstrating that, for the next few years
at least, human writers are safe as a much better source of clickbait titles than the software they claim will replace them. What is important to
Carlson, however, is that many of these stories are narratives, speculative
accounts of the possible future of the news industry, and very far from
the kind of stories that software is capable of producing at present. Also
touched upon, but largely nascent in Carlson’s article, is the fact that underlying anxiety is very much driven by the collapse of the “traditional”
model of funding media production – advertising – that we have considered previously. Automation is set to replace certain human tasks as it has
been doing on a regular basis since at least the beginning of the industrial
revolution: the cause of anxiety is whether there are enough complementary industries and activities to soak up an excess of human labour.
Carlson’s article is thoughtful in dealing with the rhetoric of automation rather than many of its actual practices, although the example of
Narrative Science draws attention to one recurring factor: the move to
niche audiences. While the tendency from the nineteenth century has
been to engage with mass audiences, the shift from print and broadcast
to digital media means that although any one report may receive few
hits, the ability to produce such reports in quantity leads to increased
web traffic in aggregate.
18
Cohen notes the tendency to refer to such output as “pink slime journalism”, after the practice in fast-food restaurants
of passing off meat byproducts – “pink slime” – as the real thing.
19
Cohen’s own case study is of a failure – a media company called Journatic
which floundered when it was revealed that it was using fake bylines
against its automated content for “journalists” in the Philippines – but
her observation that it “serves as an important harbinger for the future
of journalism” is a significant one, even if it really refers to the fast-food,
algorithmic news end of journalism.
20
She cites Bill Ryan’s argument
that cultural work such as journalism had represented the limits of the
rationalisation of labour because such practices result in a degradation
The most extensive survey of automated journalism at the time of writing is Andreas Graefe’s Guide to Automated Journalism for the Tow
Center for Digital Journalism at the Colombia University School of Journalism. Graefe considers the various means by which the use of algorithms
to gather news from highly structured sources offers an ability to produce
copy more quickly and with fewer errors than human journalists, while
also noting how it fuels those same journalists’ anxieties about their own
futures. Some of the latter issues are returned to by Carlson, who sees in
the example of the company Narrative Science a competition to automate
most if not all news content against journalists’ definitions of their own labour: very simply, software can produce a much greater number of stories,
but only when the news is defined within a strictly limited compositional
form.
16
Carlson runs through a series of headlines from a variety of publications in 2014 to 2015, such as “The Robot Journalist: Heralding an
Apocalypse for the News Industry?” (Guardian) and “Can Robots Run
the News?” (Mashable)
17
– demonstrating that, for the next few years
at least, human writers are safe as a much better source of clickbait titles than the software they claim will replace them. What is important to
Carlson, however, is that many of these stories are narratives, speculative
accounts of the possible future of the news industry, and very far from
the kind of stories that software is capable of producing at present. Also
touched upon, but largely nascent in Carlson’s article, is the fact that underlying anxiety is very much driven by the collapse of the “traditional”
model of funding media production – advertising – that we have considered previously. Automation is set to replace certain human tasks as it has
been doing on a regular basis since at least the beginning of the industrial
revolution: the cause of anxiety is whether there are enough complementary industries and activities to soak up an excess of human labour.
Carlson’s article is thoughtful in dealing with the rhetoric of automation rather than many of its actual practices, although the example of
Narrative Science draws attention to one recurring factor: the move to
niche audiences. While the tendency from the nineteenth century has
been to engage with mass audiences, the shift from print and broadcast
to digital media means that although any one report may receive few
hits, the ability to produce such reports in quantity leads to increased
web traffic in aggregate.
18
Cohen notes the tendency to refer to such output as “pink slime journalism”, after the practice in fast-food restaurants
of passing off meat byproducts – “pink slime” – as the real thing.
19
Cohen’s own case study is of a failure – a media company called Journatic
which floundered when it was revealed that it was using fake bylines
against its automated content for “journalists” in the Philippines – but
her observation that it “serves as an important harbinger for the future
of journalism” is a significant one, even if it really refers to the fast-food,
algorithmic news end of journalism.
20
She cites Bill Ryan’s argument
that cultural work such as journalism had represented the limits of the
rationalisation of labour because such practices result in a degradation