Connections

Smart Hollywood Summit: How AI, ML are Impacting Content Creation

Artificial intelligence (AI) and machine learning (ML) are already influencing content creation and, although there are still kinks to be worked out, the technology continues to evolve and improve, and it’s in the best interest of media and entertainment (M&E) companies to keep up with this dynamic trend, according to Tom Ohanian, global sales executive for IBM Watson Media & Weather.

“Ultimately,” when considering the use of AI and ML in content creation, “it comes down to these two questions: Can we produce content in a way that’s acceptable, and can we break these down to rule sets?” he said Feb. 27 during the session “Redefining Cre-AI-tivity: How AI and ML are Influencing Content Creation” at the standing-room-only Smart Hollywood Summit.

Creativity is an innately human trait, one that has rarely been considered within the realm of machine intelligence. After all, it’s difficult to imagine a world in which ideation and creative decision-making can be authentically executed by machines. Yet, the industry has made strides with sophisticated AI and ML algorithms that can mimic the work of famous painters, rewrite songs or even inform filmmaking decisions. AI can also enrich metadata so that every type of content from images, to videos, to sounds, are analyzed and catalogued to boost the value of M&E content.

“I never look at this stuff as threatening,” Ohanian said of new technologies. In the past, there’s been a lot of fear that, with new technology, lots of people will lose their jobs — but that hasn’t always been the case, he noted, pointing as an example to a plan to build connections of networks with secure, accelerated transport. That plan didn’t put people out of work, as some had feared, he said, adding: “Actually, what it did was it enabled petabytes of data to traverse over [the] Internet to completely change our supply chain, and we needed to do that,” he said. As a result of that plan, he noted that it now takes only 24 hours to deliver content from L.A. to Australia instead of a whole week.

New technology even enables people to do new and interesting things with their careers, he said, adding: “This is a journey here. There’s noting that really kind of works 100 percent to your requirements out of the box. Not yet.” But he predicted: “They’re going to get really, really smart.” There are also “pieces and, if you put the pieces together, you can build things; you can build solutions,” he said.

Among the many areas that AI and ML have been used are in speech recognition, which lends itself to closed captioning, and the creation of video highlights that are personalized, he said.

“Where it gets really interesting” is when we start “creating things that weren’t there to begin with,” he said. Although some people are afraid of that, he said: “It’s coming.” In other words, we might as well start getting used to it and even embracing it.

Moving on to the “more controversial aspects” of how AI and ML can be used, he said the question becomes: “Can you automatically create content based on rulesets and conventions?” Now, “some people get nervous about this,” but the reality is that “these things are coming [and] you’re going to see some of these things built into digital editing systems,” as just one example, he said.

Among the challenges that AI and ML specialists face is that it’s hard to get the intonation correct when editing audio, he said, but noted improvements are being made.

FCC restrictions, meanwhile, require accurate live closed captioning, he said, noting that when machines can get to a 5.1% error rate with them, that’s “on par with a human.” At that point, what can be done is to “augment systems” so that you can create an experience where AI can actually help, he said.

He pointed to the Wimbledon video highlights that IBM Watson Media created using AI. Although there were a lot of errors initially, improvements started significantly enhancing them, he said.

One advantage of using AI and ML more is that the technology can be used in “easing human workloads,” so that people can do other things that are more challenging than repetitive tasks, according to Ohanian.

Investing in R&D, meanwhile, is challenging, he conceded. After all, in the M&E sector, “there are 3,00 vendors” and 85 percent of them “have less than three products,” and some of those vendors are “really scraping to get by,” he said. At a typical NAB show, 28% of the companies are up for sale, he claimed. Therefore, while investing in R&D, a company doesn’t always know if its vendors will still be in business in the near future, he said, noting that only about 52% of Fortune 500 companies that existed 15 years ago are still here.

This year’s Smart Hollywood Summit was produced by the Media & Entertainment Services Alliance’s Smart Content Council with sponsorship by IBM Watson, MarkLogic, EIDR, Hammerspace, human-I-T, Independent Security Evaluators (ISE), KlarisIP, Testronic, FilmTrack, OnPrem, Mediamorph, RSG Media, Vistex, Vubiquity and Bob Gold & Associates.

To download the audio of Ohanian’s presentation, click here. To download the PDF of his presentation, click here.