Page 95 - AC/E Digital Culture Annual Report

and possibly for others.” Nevertheless, one hopes that with time and effort, the results improve.
It is important to realize that the work presented to NaNoGenMo is not usually in the shape of novels in the traditional sense, but attempts to explore new novel concepts quite separate from those usually on shop shelves. This confirms the idea that even in the case that these kinds of initiatives should reach a reasonably good level, it would be strange for them to compete in the same market as books written by humans. How- ever, there is a chance that during the process
of investigation, new interesting formats are discovered that perhaps could later be adopted by human authors – who could possibly improve the quality. Another possibility is that they come up with interesting formats that could not have been invented by a human, but which are easy for a machine.
Service Companies
Another current phenomenon suggesting where, in the production chain, AI technology could start to bring added value is the offer of auto- matic text generation services. The company Narrative Science7 offers automatic data process- ing services that convert the data into intelligent narrative. For a collection of numerical data on
a concrete subject such as a baseball game or a company’s accounts, these services are capable of generating a small story that tells the most important events or the most important links be- tween the received data in accessible language. Narrative Science’s main tool is a program of text generation called Quill. The company Automated Insights8 offers similar services. The company’s advertising declares that it automatically gener- ates five hundred thousand million stories of this kind a year for companies all over the world. It uses the Wordsmith tool that works on masses of data to automatically generate narratives that sound as though someone had written each one individually. In both cases, the companies insist that they are not trying to replace the human
writers who previously did this kind of work. The quality of the generated texts is still inferior to those written by humans.
Automatic text allows, without incurring cost, what would be prohibitively expensive if done by human copywriters.
The companies that can afford to take on human copywriters or those areas that deserve to be covered by specialized journalists, will continue to receive the attention of professional copywriters. The automatic word processing
will not replace them, but will rather allow other clients who have previously not received written attention to use this technology to generate their stories. It is also emphasized that the technology allows the stories to be personalized to fit the user’s particular circumstances. There is not just one version of each game or economic circumstance. Instead, there is a particular story generated around the events according to the interests of the client and each story is very specifically adapted to the needs of the client. Automatic text allows, without incurring cost, what would be prohibitively expensive if done by human copywriters.
Conclusions
The analysis offered in this article demonstrates that the general fear that computers will usurp writers’ jobs in the medium term is completely unfounded. The developments in AI technology dedicated to this area are focused on under- standing the phenomenon of creation at a fundamental level, beginning by breaking down the simplest processes that they are trying to imitate. The moment when all the simple pro- cesses involved in literary creation are modeled is still a long way off. Literary creation is a very complex task that requires complete models of the human mind, including cognitive capacities such as the ability to recall, relive experiences of all kinds, feel empathy with others or understand complex emotions. All this is still very much
AC/E DIGITAL CULTURE ANNUAL REPORT 2018
95
Digital Trends in Culture