Artificial intelligence (AI) tends to split opinions when it comes to creative mediums, with traditionalists erring away from such a seemingly hands-off technology – but could it actually make life easier for those in the scripted game? Stephen Arnell finds out.
As in so many areas of life, AI is encroaching into the world of scripted television. But what, in this context is AI?
Not a vaguely sinister, sentient HAL 9000-style super-computer as envisioned by low-tech types (myself included), but, in the more prosaic words of Swiss company Largo.ai, online system software that “enables an understanding of the ingredients of a film/pilot and predicts the audience reaction and potential revenue, from script level through to rough and fine cut.”
It goes on: “Through a cognitive pattern understanding system it creates a ‘recipe of ingredients’ together with actionable insights.” Clear?
The film world has already embraced the use of AI. Back in January 2020, Warner Bros. inked a deal with CineLytic, an AI-driven business intelligence firm, to provide predictive analytics to aid decisions prior to greenlighting, assessing the value of star names, forecasting revenue both from cinema releases and ancillary streams.
CineLytic is currently focused on film, but aims to expand into episodic content before year-end. Its founder Tobias Queisser explains: “We identified which key variables of a film project drive performance and built a forecasting tool that can essentially prototype content and allows the user to run various forecasting scenarios based on key variables including genre, talent, rating, IP, release timing and distributor etc.
“The platform reduces executives’ time spent on low-value, repetitive tasks and instead focuses on generating actionable insights for packaging, green-lighting/financing, marketing and distribution decisions in real time.”
Talent selection & plot advice
Sami Arpa, Largo.ai’s CEO and co-founder, claims to have advanced beyond competitors in “performing analytics before and during the production of the film/TV production rather than after release.”
This way, he says, “we can increase the size of the market for film and TV with insights for selecting and optimising the content mix.”
As the so-called ‘golden age’ of TV and streaming drama continues, AI could be a useful tool in helping commissioners separate the wheat from the chaff in terms of the daunting volume of prospective projects. Looking towards streaming applications, Arpa highlights that “both Netflix & Amazon initially started to use AI as recommendation systems. Later with their original content, they aligned production and acquisitions with AI outputs and learning.”
He explains that Largo.ai expects attitudes and uptake to shift within the TV industry to level out the playing field, with the pandemic significantly increasing demand for innovative TV and streaming content. Arpa adds that his service is able to make streaming predictions using pilot episode scripts, identifying key character relationships, for example, or proposing matching talents and “possibly even beneficial plot changes.”
This, Arpa asserts, means that the company is in a position to provide content insights, “giving new perspectives and in-depth character analysis, which matches the most suitable actors for each role. Since users define their budgets, the system can suggest the most fiscally appropriate actors.”
Quite what the Casting Directors Guild UK/Ireland and their equivalents around the globe think of this prospect has yet to be disclosed, though Arpa says that Largo.ai has had very little in the way of creative pushback. “We’ve had a very positive and encouraging response from producers.”
The human factor
Movie marketing company Movio, meanwhile, focuses on audience behaviour to understand how past patterns of movie-watching can help to predict someone’s likelihood to watch in the future.
Its chief client officer, Sarah Lewthwaite, offers insight that could just as easily apply to scripted TV. “We can identify macro trends and opportunities,” she says. “For instance, AI has signalled an opening to target women aged 30-50 and to expand US movie attendance by introducing consumers to cinemas via more ethnically diverse stories.”
Of course, in the back of some executives’ minds could be the fear that AI may reject a show before they see it. It could then go to a rival – and become a huge hit. Overcoming such suspicions in the creative community is a battle,” says Paul Youngbluth, joint-MD of veteran non-AI TV consultants Tape. “Your conclusions must be delivered in a positive and non-confrontational manner, sympathetic to the idea that creativity is a unique process,” he explains.
Youngbluth further suggests that a human element still needs to exist in the process. “It is useful as a predictive tool – encapsulating a huge amount of information and extrapolating likely outcomes. But AI-led conclusions need to be re-interpreted by a human process (the writer) and presented in such a way that inspires creativity rather than attempt to replace it.”
He also points out that the likely positives in using AI will mean escaping some of the drudgery of producing scripted TV.
“In many ways, forecasting ratings is a process best suited to an AI approach, with modelling, algorithms and stats all helping determine the ultimate numerical outcomes. AI can also help determine the likely demographic appeal of a new project, but be aware that small target groups may be given undue prominence.”
Dan Korn, producer and channels industry veteran, meanwhile describes the use of AI as simply the next iteration of gathering data about audience likes and dislikes. “Whilst we will always need the human element as the ultimate arbiter in the process of bringing shows to the screen, AI could offer insights that might not otherwise have been available.”
Korn does, however, caution that “creative risk, self-expression and serendipity” has always been part of the process. “The idea that ever-more sophisticated AI could create homogenised ‘product’ that eliminates the ‘happy accident’ or left-field creative decision is a concern, as is the possibility that its predictive ability could be dependent on data and outmoded assumptions, which could swiftly become redundant.”
Speaking to Korn’s point regarding the importance of up to the minute data, prior to season eight of HBO fantasy hit Game Of Thrones, prospects looked bright for spin-offs. Now? Well, perhaps not so much.
AI as data assistance
What’s clear, however, is that the use of AI is increasingly being incorporated in the scripted business. Sunder Aaron, co-founder and MD of popular streamer The Q India, explains how he has developed projects using AI data assistance from Parrot in the past.
“This helped us to successfully prepare the pitch for platforms. We also provided the data results to potential partners, which I believe strengthened our pitch.”
He does not, however, believe the technology is yet ready to impact the actual creative process. “At this stage, AI tools can be more effectively used for general decisions about genre, casting, title and demand estimation, but not currently for the writers room or making production decisions.”
Understandably, broadcasters may be reluctant to discuss their current or planned use of AI technology in helping to decide what scripted projects to pursue. And creatives may well blanche at the thought that their cherished scripts are being evaluated by a non-human intelligence.
Indeed, in 2016, the producers of AI-scripted horror feature Impossible Things sought crowdfunding for the project, but, as of May 2021, it still is yet to be made. And a warning from history in the fate of Relativity Media, whose data-driven Moneyball approach to their movie slate apparently contributed to the company declaring bankruptcy in 2015.
CineLytic’s Queisser has sought to quell fears, reassuring those frit by the prospect of an Ultron-esque overseer, (as seen with the villain of Marvel movie Avengers: Age Of Ultron). “Artificial intelligence can sound scary. Yet right now, AI is very limited when it comes to creative content decision making.
“What AI is good for at the moment is crunching numbers and breaking down huge data sets to recognise patterns for content business decisions that would not be visible to humans. For creative decision-making, you still need experience and gut instinct. Combining data-driven AI and human creative thinking can lead to strong synergies.”
“We always stress that Largo.ai is an assistance tool,” agrees Arpa. “It enables rather than replaces the creative process by underpinning human instincts.”