What’s in store for the future of synthetic media?
While many people may not be familiar with the term “synthetic media,” there’s a growing concern around the way the technology can be used. More commonly known as “deep fakes,” synthetic media isn’t limited to manipulated images or video but can encompass any form of media that is generated or modified through algorithmic means.
Examples can range in complexity from a person’s image or voice being manipulated to complete personalized digital experiences being created from scratch. This is usually done through some form of generative adversarial network (GAN) or style transfer technique.
Many experts are concerned about the use of synthetic media for fraud and manipulation, but the technology also has the potential to upend the media landscape. It can do so by enabling the creation of high-quality, personalized content at scale both faster and with fewer resources.
To gain a better understanding of why synthetic media has garnered so much attention and where it is headed, we have identified both the challenges it faces and the opportunities it presents for those who embrace the technology.
The result of this research is our first-ever synthetic media landscape, which breaks down use cases and profiles 50 startups in the space. You can download the landscape at syntheticmedialandscape.com.
Moving past deep fakes to widespread potential
Most references to synthetic media today are related to how technology can be used to manipulate existing content in a fraudulent or misleading way.
As the United States moves closer to its election in November, for instance, there are growing concerns that individual campaigns or outside actors could weaponize deep fakes or manipulated media for an edge. The acceleration of media cycles and speed of distribution on social media could lead to the doctored video of a candidate going viral with little room for fact-checking.
By focusing on fraud, however, many people overlook the vast potential synthetic media has. In the hands of non-malicious actors, the technology can be used to create new kinds of content and services — and it can do so much faster and at a much greater scale than ever before.
Using synthetic media to overcome barriers to content creation
The primary challenge to creating content comes from the limitations of valuable resources like time, cost, and overall effort.
If you’re creating an ad, for example, you have to take into account items like production, location, and staffing, which can add up very quickly. You also might want to hire top, recognizable talent to endorse your product. That talent is not always available because they can only be in so many places at once. They’re only human, after all.
With the right data, technology, and computing power, however, you can now create new content that costs significantly less than before. You will also have the ability to personalize content for the individual, which creates a stronger impact on the viewer and improves their experience.
Similar models could be used to create entertainment experiences like TV shows, games, or even articles — all of which have seen an increase in demand during COVID. Being able to produce engaging content without hiring talent or having people on a set could be a serious advantage during these times.
There are already examples of photorealistic, personalized content being produced at scale. For example, companies like Hour One are working with clients to generate synthetic characters that are based on real people, all in the goal of increasing sales and engagement.
Improving customer service and experience
Synthetic media could also have a massive impact on customer service. We have all grown used to interacting with bots over the past few years, but we deal with it because it can often get us to the information that we need faster than waiting for an available agent to answer the phone.
Synthetic media could vastly improve the customer experience and make it feel more personable by using a realistic avatar to mimic human interactions. Alta Voce, for example, shapes voice in real-time to improve communication with support agents.
Of course, these advancements are not without their challenges, especially as they pertain to privacy concerns and regulations.
Overcoming obstacles to personalization and privacy
Providing a personalized experience depends on the ability to collect and process personal data. However, we are starting to see pushback against this type of data collection. As surveillance grows, consumers and governments are pressing for stronger protections. The demand for more privacy in turn makes personalization more challenging.
A clear example of the challenge is in the collection of video footage by retailers at their stores. Whereas cameras may have initially been installed for security purposes, they now help to provide businesses with information about consumer behavior and to offer specific customers incentives like coupons to encourage sales.
Under certain regulations, these forms of data collection can land retailers in hot water. At the same time, some companies have invested serious resources into technologies that can and should help them to improve sales. Thankfully, this is another area where synthetic media can have a leg up over existing models.
One startup working in this field is D-ID. Its technology processes video content so it cannot be used for facial identification. This pseudonymization allows retailers to still reap most of the business benefits without identifying specific clients.
Technological advances in synthetic media
Market demands play a part in the adoption of synthetic media but most of the credit should go to a few advancements in technology that make it financially viable at scale.
The first two areas of significant improvement have been in Generative Adversarial Networks (GAN) and style transfer. These are the techniques that allow creators to produce synthetic media that can automatically respond to a user. Improved cloud availability is already a factor in helping to improve content creators’ ability to utilize GAN and style transfer.
There are challenges that will need to be addressed through innovation when it comes to increasing computing power, however, which includes decreasing the datasets needed for personalization and optimizing costs. Like the need for advances in anonymization, finding solutions to these challenges will create big opportunities for companies willing to take up the challenge.
Our role in the future of synthetic media
As an innovation group, we want to provide more visibility into this growing industry. As such, our team has created a map of the synthetic media landscape and hopes to raise more awareness around the industry.
We invite you to review the map and join the discussion. Please send me your thoughts at Royi@samsungnext.com and reach out if you know a company we have missed. Hopefully, this map will be a useful tool for those considering opening a venture in this space or investigating it.
Originally published at https://samsungnext.com on August 11, 2020.
