Creative or primitive: AI breakes into movie industry

Креатив или примитив: ИИ проникает в киноиндустрию Креатив чи примітив: ШІ проникає до кіноіндустрії Creative or primitive: AI breakes into movie industry

Artificial intelligence set its sights on movie industry. Is there any place for spiritless machine among bohemia?

In 2016, McCann Advertising Agency astonished the entire world with the news: artificial intelligence (AI) became the Creative Director of its office in Japan. A robot, created by one of the employees, won a competition with a human. It has written the advertising campaign brief, where it defined the campaign goal, target audience, claim the advertising should make, and key advertising elements. Promotional video created based on this brief was shown at the UK advertiser trade body ISBA’s annual conference in London. And this video was much more better than the one, created by humans.

Promotional video created by robot:

Promotional video created by humans:

Has AI already become creative? No, not really, in fact. It will never handle the creative task without the help of a person. The teach the robot of how to generate ideas, employees of McCann Advertising Agency uploaded to it database of deconstructed advertisements from all the winners of some of Japan’s biggest award shows from the past 10 years. Moreover, people were dealing with promotional video production by themselves.

Креатив или примитив: ИИ проникает в киноиндустрию Креатив чи примітив: ШІ проникає до кіноіндустрії Creative or primitive: AI breakes into movie industry

Robot-creative specialist from McCann Advertising Agency

Regardless the fact that artificial intelligence is unable to be creative, it is capable of some things.

The first steps

Researchers from the Massachusetts Institute of Technology (MIT) have made AI to think creative somehow. When a person looks at a picture of moving train, he may predict which way it is moving and what will happen at the next picture. This is a challenging task for artificial intelligence, but researchers are close to finding a solution. They trained AI to turn images into videos and predict what happens next.

That very MIT has developed yet another algorithm able to create movies from text.  Two neural networks are used for this purpose. The first one creates image and the second one improves it. The results of the study are published in Science journal.

Movie trailers creation is another sort of work that may be performed by artificial intelligence. Thus, IBM’s AI Watson created trailer for the Morgan movie. To achieve this goal, neural network was fed with more than one hundred horror movie trailers. All the process lasted 24 hours, while if this job is performed by a group of people, it will take more than a month. In other words, AI allows us to save a lot of time and may considerably simplify content production process. Selection, analysis and adaptation of materials needed for the movie industry – all relatively minor tasks may be entrusted to AI.

But that is the future. So far, “creative activities” of AI are rather primitive. “All this is just a big start, – Viktoriya Tigipko, Founding Partner at TA Ventures venture capital fund and President at Odesa International Film Festival, said. – On a positive note, in any case, there will always be work for those, who think outside the box.

AI “sees” the future of a movie

Media and production companies are currently using AI to resolve tasks that accompany production – search for materials or advertising optimization.

Thus, in June 2016, Megogo online cinema started testing of neural network that analyzes the content and shows the relevant advertisement to the viewer. For instance, if on the fifteenth minute of the movie some beverage appears in the frame, neural network will immediately show advertisement of this beverage. Ivan Shestakov, Head of Marketing and Strategy at Megogo, believes that neural network will find any pattern, if it is properly taught. Anyone may deal with teaching – all it takes is to feed the program with 500 images of an object and it will choose patterns by itself. When this is done, a person shall confirm the correctness of this choice. “The longer the teaching process is, the more accurate the result is delivered,” – told Ivan Shestakov in his interview to Innovation House.

But this technology failed to move ahead of testing phase. According to Mr. Shestakov, Ukrainian advertisers are mostly interested in audience coverage, rather than striking the target. For this very reason, so far, innovation proposed by Megogo is non-demanded. “There were several commercial projects, but it would be premature to expect these projects to be extensive,” – Ivan Shestakov explained.

1+1 Media Group is also preparing the ground for the use of artificial intelligence. Digital Department of the company is currently beginning its work with big data, since it is the primary source of AI solutions. Data model shall be defined, described and compared to various processes. Once this is done, we will be able to talk about development of required generative model, for instance.

“AI implementation is rather expensive and resource-intensive process. Nevertheless, we are gradually we are moving towards it,” – Head of Digital at 1+1 Media Group said.

Prime tasks that 1+1 Media Group hopes to solve using AI – improvement of content recommendation system on all informative and entertainment venues of the group, to “breath life” into existing chat bots, to develop SUDUM copyright protection system, and to generate analytical articles.

Each of these tasks needs better and greater understanding of what viewers wish. This is a large amount of data that shall be stored, described, processed, and only then predictive analysis of consumption may be developed.

There are startups that are already creating predictive analytics for movie and media industry. Scriptbook from Belgium, for instance. It predicts box office results and viewer rating based on one scenario only. Artificial intelligence is analyzing the database of movies since 1970 and IMDB ratings to make forecast. Scriptbook has predicted that Passengers movie that was made in 2016 will generate $118 million revenue at the box office and will get 7.3 IMDB rating.

Israeli startup Vault ML is also using the algorithms of artificial intelligence. It analyzes the database of viewers and provides recommendations on when the movie premiere shall be made. Vault ML has already raised $1.4 million of investment, while the total amount of investment in Scriptbook is $1.2 million.