Storytelling in the Age of Man-Machine Collaboration
Early efforts at AI generated sci-fi screenplays may be laughable, as evident from the unintelligible dialog, lack of a discernible plot and vague attempts at humor in “Sunspring” a short film written by an AI bot. But the future of collaborative storytelling with machines is rich with possibilities if creators and engineers find innovative ways to work together. This is already happening in many news and media organizations, and as a filmmaker and content creator I find the creative possibilities very exciting. However, the more powerful opportunity is to apply new technologies to bridging the gap between stories being told and the very important ones that are untold because of lack of access to the storytelling engine as well as limitations of language and logistics.
Not only is AI capable of sifting through massive amounts of data to detect pop culture trends, or fraudulent activity, machine learning is capable of creating smart content and connecting it to finely targeted audiences. Twitter is now using AI to analyze tweets to predict a riot sooner than police, and recently The International Consortium of Investigative Journalists (ICIJ) parsed 2.6TB of data in 11.5 million documents to expose tax havens being exploited by the rich and powerful politicians in what has come to be known as the Panama Papers.
AI, even in these early days, is a powerful tool for creation. Natural Language Generation (NLG) AI is being used to auto-generate stories based on pre-determined templates. One can train an AI to interpret large amounts of reports, events, corporate financials etc., and leverage advances in machine vision to search massive amounts of visual data and create video compilations. Imagine this kind of method being applied to the BBC archives of unused news footage, where a well-designed algorithm might come up with a surprising new narrative for the JFK assassination, 9/11, or Tiananmen Square. In fact, AI news editors and producers might present alternative, rigidly accurate interpretations of current events shorn of the editorial biases and slants of their human counterparts.
A collaboration between MIT Media Lab and McKinsey and Company recently found that it is possible to discern common emotional arcs in video stories and the characteristics of these arcs are directly related to audience engagement. In fact, the project went on to analyze slices of video (TV, movies, shorts etc.) to estimate positive or negative content by the second. AI can also view thousands of films and shorts and break them down into categories of emotional arcs. Apply to this mix, the type of data-driven work Buzzfeed and others are doing to increase the virality and stickiness of their content, and soon we might know exactly what content to serve to a specific audience to gain viewership and engagement. All this and more, and we haven’t even started applying AI to boost artfulness and create new types of story formats.
But as technologist and entrepreneur Ali Khoshgozaran points out, perhaps the most high-impact opportunities for applying AI and machine learning would be to eliminate the digital divide and illiteracy, correct the inequity in access to knowledge and learning, and bridge the technological gap to bring the world closer. He mentions Google’s translate service and Lingua, its human-powered equivalent, as well as Skype’s real-time translator as examples of successful applications of AI in storytelling for social good.
For storytellers, the possibilities arising from man-machine collaboration are endless, and there exists a powerful opportunity to bring unheard voices and narratives into the mix and generate new mythologies for the future.
“Sunspring”, a Sci-fi short starring Thomas Middleditch, was written by an AI bot