Just Knowin: AI Can Improve Content Creation and Enjoyment
Recently, we had lunch with Andy Cochrane to talk about a series of sessions at being planned for Fall’s FMS (Flash Memory Summit) on the state of VR, immersive and holographic film work (they’re going to be exciting, informative; trust us).
Our discussion drifted to the speed of technology change taking place in the M&E industry.
Cochrane – who’s a great Indie content maker – jumped in and said, “Yeah, isn’t it exciting!”
Okay, we’re not sure why filmmakers seem relish uncertainty, but they always seem to be up to the challenge.
Our biggest issue with the newest hot techie topic is the name … artificial intelligence.
It’s so … artificial.
We were delighted when Ginni Rometty, IBM’s president/CEO, keynoted at CES describing AI as more of an augmented or assisted technology rather than artificial … assist people’s intelligence, not replace it.
She also warned it had to be released into the world slowly after a lot of thinking and testing because you can’t put it back in the bottle once it’s out there.
Think back on 2001: A Space Odyssey, AI: Artificial Intelligence and the other forecast-of-the-future films.
Think about how many far out, historical films have been a roadmap to tomorrow.
But still, Cochrane said many of the ways AI could used in filmmaking would be tremendously useful.
“You have to understand that a lot of a filmmaker’s work is tedious and time-consuming,” Cochrane noted, “especially in the pre-production phases – script breakdowns, storyboards, shot list generation, schedule optimization and budgets. Once we streamline these activities, we can focus on the creative work.
“Those mundane, routine tasks can be streamlined and probably done more accurately by ‘trained’ systems,” he said. “That frees up filmmakers to focus on developing films/shows that are more interesting, more involved.
Think of it,” he commented, “We could then create projects that are more engaging, immersive for the viewer; and, more importantly, we can have a greater assurance they will be profitable.”
He really does make AI does sound mystical, magical.
Or, as futurist Arthur Clarke deciphered the idea in his third Clarke’s Law – “Any sufficiently advanced technology is indistinguishable from magic.”
The development and use of GANs (generative adversarial networks) or networks vs. networks is one application that reinforced Ms. Rometty’s concern about rushing AI applications to market.
Researchers around the globe have used GANs to train AI algorithms.
For example, Nvidia researchers used GANs as a “style-based generator” to create hyper-realistic images.
Since then, GANs and the style-based generators have been used to produce fake political videos, create/distribute extremely embarrassing/compromising content that looks so real it is difficult to explain or dispute.
Fortunately, Cochrane only associates with people who have integrity. “It’s better to hang out with people better than you, he said. “Surround yourself with people who are better than you and you’ll drift in that direction.
“It’s a question of ethics and professionalism,” Cochrane continued. “Filmmakers I know well and work with really just want to create a film that will transport people to another part of the world, educate and entertain them. We’re using the best tools, people and techniques to deliver a great product to viewers.
“When it’s a wrap, the hard work begins;” he added, “getting it sold and seen.”
In today’s crowded theatrical, appointment TV and OTT marketplace it’s an even trickier, more dangerous task.
In many ways, AI is enhancing filmmakers’ creativity, not squelching it.
For example, investors have shown a lot of interest in Scriptbook which claims that its algorithm is three times better at predicting box office success than human readers.
The company has spent a lot of time/effort analyzing scripts of completed successful and not-so- successful films and projects to say, “See it works. The analysis shows that you wasted a ton of money on those three projects that bombed; or yep, our system ‘validated’ that this was going to be good, great, dynamite!”
We do know we didn’t see the buyers at Sundance sit through any of the screenings with systems on their laps doing a bunch of number crunching before they wrote checks.
They looked at the team involved in the project, flipped through the storyline, studied the pre-event buzz, checked the audience response during the 10-day Park City event and did deals.
Not that AI won’t be involved in content projects at the beginning, because the industry has accumulated a lot of data.
Movie data includes box office revenues, production costs, and audience demographics as well as director, screenplay writer, cinematographer, and other detailed project information.
So, we can expect intelligent systems to find their way into the industry to dramatically transform how – and if – stories are told … and sold.
Legendary Entertainment and several competitors have put their AI tools to work to develop movie, show and series trailers to optimize audience interest.
Qloo has been busy with their AI solutions developing viewer and audience profiles to target marketing efforts and activities to create measurable positive results.
Products/services like Qloo are also being used to identify viewer preferences, likes and dislikes so content developers/distributors can determine which projects get funded when they don’t want to rely on their experience and … gut.
Fortunately, most film festival entries don’t get produced by committee or AI but by folks who believe in the story that they develop.
Studios, networks and streamers invested quickly and heavily because they had a pretty good idea how to use subscriber/viewer data to attract the right audience.
All of the major streaming services – Netflix, Amazon Prime, Hulu, YouTube, TenCent – have an overabundance of viewership data to determine which content will do well; and, more importantly, how to tailor their trailers and social media efforts to attract viewers and encourage them to tell others about the projects to subscribe.
The OTT services base their content projects, purchases and airings based on the content people have watched, when they watched it, on the screen used and more individualized data to customize how the video content is pitched to you.
“Some people view this data capture and mining as a bad thing,” Allan McLennan, head of PADEM Media Group, a leader in OTT as well as a pioneer in AI, VR and iTV commented.
“These services constantly run A/B tests, analyze when people hit play/pause, when they stop watching a title, add something to their watchlist and analyze other data to optimize an individual’s enjoyment,” he noted. “The data and analysis allow them to more accurately personalize the service for the viewers.
“As their use of AI solutions becomes more sophisticated and common, it will enable producers and programmers to identify what new projects should be developed and even to some degree the viewership/engagement that will enjoy their production before it begins,” he added.
“The technology doesn’t control or mold what people want to watch” McLennan emphasized. “And it certainly doesn’t control whether good or bad video stories will be made. That is still a very personal, human activity.”
Cochrane and McLennan agree that most of the “noise” we see today regarding AI is headline-grabbing stuff – widespread job elimination and control/manipulation of what people think, feel, do.
Cochrane feels that the technology will enable filmmakers to streamline note taking and project management as well as helping to create projects that people can connect with and react to more quickly.
“Filmmaking is–and will continue to be–a very personal endeavor,” Cochrane emphasized, “and the new tools will make it faster, easier and probably more profitable to get viewers to connect with that story, that message.”
“AI has a lot of potential for every segment of the M&E industry,” McLennan said, “from concept and delivery to engagement.”
“One of the most important is streamling and improving communications throughout the creative/production process,” he added. “In fact, as simple as it may seem, one that could extend off of AI/metadata intelligence, and I’m sure is on the horizon, is one that could help all of us manage all of our editing processes, even our email. We all know how much of an assistance that would be.”
AI holds a lot of potential for the M&E industry and should be viewed as an opportunity to improve the personalized experience rather than a threat to content creators. People tend to be limited by what they know today, not what we imagine might be true tomorrow.
# # #