Turing’s Test 119
Eyre’s story is his prose style: as was made clear earlier in this chapter,
there is a brand of Anglo-American reporting that has effectively created a “news algorithm”, and indeed even in the comparison of hospital
cancellations produced by an algorithm and a person in the previous section, there is frequently little difference in the style of writing. However,
a virtue of a series of rules – which Eyre frequently abides by to provide
his story with greater authority – is that these can then be broken for the
greatest effect. This, indeed, is what Eyre does in the first sentence of his
story: rather than providing the who, why, what, where, when, and how
of his data, he employs a narrative hook used by a great many writers –
though more brutally and more effectively than most, emphasising the
moral consequences and judgements of the story he will tell. In lesser
hands, this would be editorialising, but Eyre’s strength is then to hold
those judgements in reserve, to allow his data to inform the reader.
As the entire article unfolds, Eyre interweaves other elements that are
currently impossible for software to provide: interviews with authorities and those on the receiving end of prescription opioids. There is no
algorithm at present that has demonstrated itself remotely capable of
sourcing quotations and stories from interviewees, but there are also
fundamental problems with language processing that create extremely
high barriers for the extension of algorithmic journalism beyond certain
limited fields. In 2002, Robert Cameron described the four concepts of
natural and computer languages as syntax, semantics, pragmatics, and
metalanguage, and while many developments have forged ahead in the
past 15 years with regard to semantics and syntax (what words mean
and how they are arranged together into meaningful sentences), dealing
with pragmatics (the context of messages) has often proved more difficult. While textbooks such as Programming Language Pragmatics address in great detail the issues surrounding the interpretation of syntax
trees as a basis for understanding the contexts of language, such things
as creating variables for environmental references assume that these can
be easily quantified.
72
When it comes to programming Alexa or Siri to
understand that “play next” refers to a song on your Spotify list, or that
“turn off the lights” can be set to a particular circuit of bulbs in the living room, algorithms are improving all the time. Creating the emotional
and intellectual variables that will understand the anger, pain, humiliation, and fear of dealing with opioid addiction – or the recalcitrance,
evasion, and lies of those who wish to cover their tracks – like many
things appears to be currently beyond the capabilities of NLP.
This is not to make a glibly pro-humanist and anti-software statement, nor is it to assert that emotional intelligence (or at least aspects of
it) can never be programmed. After all, in many respects, it is precisely
his education at Loyal University of New Orleans and his experience on
the Gazette-Mail that “programmed” Eric Eyre to be able to spot a story
and craft it into such a compelling series of articles. Yet as was noted
Eyre’s story is his prose style: as was made clear earlier in this chapter,
there is a brand of Anglo-American reporting that has effectively created a “news algorithm”, and indeed even in the comparison of hospital
cancellations produced by an algorithm and a person in the previous section, there is frequently little difference in the style of writing. However,
a virtue of a series of rules – which Eyre frequently abides by to provide
his story with greater authority – is that these can then be broken for the
greatest effect. This, indeed, is what Eyre does in the first sentence of his
story: rather than providing the who, why, what, where, when, and how
of his data, he employs a narrative hook used by a great many writers –
though more brutally and more effectively than most, emphasising the
moral consequences and judgements of the story he will tell. In lesser
hands, this would be editorialising, but Eyre’s strength is then to hold
those judgements in reserve, to allow his data to inform the reader.
As the entire article unfolds, Eyre interweaves other elements that are
currently impossible for software to provide: interviews with authorities and those on the receiving end of prescription opioids. There is no
algorithm at present that has demonstrated itself remotely capable of
sourcing quotations and stories from interviewees, but there are also
fundamental problems with language processing that create extremely
high barriers for the extension of algorithmic journalism beyond certain
limited fields. In 2002, Robert Cameron described the four concepts of
natural and computer languages as syntax, semantics, pragmatics, and
metalanguage, and while many developments have forged ahead in the
past 15 years with regard to semantics and syntax (what words mean
and how they are arranged together into meaningful sentences), dealing
with pragmatics (the context of messages) has often proved more difficult. While textbooks such as Programming Language Pragmatics address in great detail the issues surrounding the interpretation of syntax
trees as a basis for understanding the contexts of language, such things
as creating variables for environmental references assume that these can
be easily quantified.
72
When it comes to programming Alexa or Siri to
understand that “play next” refers to a song on your Spotify list, or that
“turn off the lights” can be set to a particular circuit of bulbs in the living room, algorithms are improving all the time. Creating the emotional
and intellectual variables that will understand the anger, pain, humiliation, and fear of dealing with opioid addiction – or the recalcitrance,
evasion, and lies of those who wish to cover their tracks – like many
things appears to be currently beyond the capabilities of NLP.
This is not to make a glibly pro-humanist and anti-software statement, nor is it to assert that emotional intelligence (or at least aspects of
it) can never be programmed. After all, in many respects, it is precisely
his education at Loyal University of New Orleans and his experience on
the Gazette-Mail that “programmed” Eric Eyre to be able to spot a story
and craft it into such a compelling series of articles. Yet as was noted