The case for computers, creativity and natural language generationAdd bookmark
As computer intelligence grows, so too does the question of who holds the creative straw—man or machine?
Photo by Aaron Burden on Unsplash
Would you read a book of poems authored by a machine? It’s a question that divided the Chinese literati when The Sunlight that Lost the Glass Window, a book of 139 poems written by one intelligent bot, was authored in May this year.
In a similar case 12 months earlier, scholars were perplexed by the decidedly mournful poetry generated by Google’s artificial intelligence after it was fed more than 11,000 unpublished romantic novels.
Whereas they were once relegated to the science-fiction shelf, cases such as these are becoming increasingly common as natural language intelligence progresses in leaps and bounds.
In particular, the development of natural language generation (NLG) systems, which are capable of translating computer data into human language, is redefining the way computers and humans interact.
Once used primarily by weather bureaus to transfer large amounts of technical information, NLG is now being applied to a wide variety of sectors—think everything from financial services and government to retail and even the arts.
Would you read a book of poems written by a machine? Photo by Thought Catalog on Unsplash
“Written language is no longer a uniquely human construct,” said Leah Henrickson, a doctoral student at Loughborough University who is currently conducting research into the social and literary implications of NLG. “We’re now at a point where computers generate texts largely indistinguishable from human-authored texts, and at rates incomparable to that of humans.”
According to Henrickson, this progress brings a plethora of unanswered questions to the fore. Questions like, “what happens when we're not the only ones writing texts?” and, “how are we supposed to interact with computer-generated content when computers don't (at least right now) seem to have the lived experiences and self-awareness that we assume human authors have?”
Bot or not?
The biggest question of all—that of how far machine intelligence is likely to encroach on more creative sectors—is the one that hangs heaviest. There are many who look to computational creativity, the field which directly negotiates the relationship between computers and creativity, to answer this.
“Basically, it’s a question of whether computers are mere tools for creative work, or whether they are creative agents in and of themselves,” said Henrickson. And it’s a question that fascinated Australian writer and poet Oscar Schwartz so much that he is undertaking a Phd into whether computers can in fact write poetry.
In his TED Talk, Schwartz explains how his research led to the development of the Turing test for poetry—a test aimed at understanding whether humans can tell the difference between poetry written by man or machine. Computer scientist Alan Turing, who developed the Turing test in 1950, did so with one question in mind: can machines think?
According to the test, if a computer is able to make a human think that it is also human more than 30 per cent of the time, then it is considered an intelligent machine. By applying the same principle to poetry, Schwartz’s research engine—aptly named bot or not—found instances where poems generated by computers have convinced as much as 65 per cent of respondents that a human wrote it.
Schwartz’s conclusion is a simple one: that humans are “not a scientific fact” but rather an “ever-shifting, concatenating idea” changing over time. “We shouldn’t only be asking ourselves, ‘Can we build it?’ But we should also be asking ourselves, ‘What idea of the human do we want to have reflected back to us?’”
From bard to boardroom
While the question of computers and creativity won’t be solved anytime soon, the business case for NLG technology is a strong one. Simply put, it’s about aiding people in automating tasks that are data-driven or require a standardization that mostly eliminates creative freedoms, said Adam Long, VP of product at Automated Insights, the US-based company behind the NLG platform Wordsmith.
The promise of this technology, Long said, is threefold; more targeted ways for companies to communicate with customers; less inefficiency where manual tasks can be automated; and decreased time between insight and action.
Want more? Try this: Don't fear superintelligent AI with Grady Booch
“By automating narrative generation with NLG technology, like our Wordsmith platform, it frees up [human] time for more creative writing and tasks,” he said. Those currently benefiting most from this automation include journalists, data analysts and decision makers who are beholden to complex dashboards and cumbersome data analysis.
No matter the sector, it’s in working together with machines that humans stand to gain the most, Long said. “By pairing humans and software together, you can produce something that’s much better than either one can do on its own.”
Become a member of the AI & Intelligent Automation Network for free access to industry news, whitepapers and articles. Sign up today.