Yes, positive deepfake examples exist



Artificial intelligence technology, to many, is scary. It’s packed full of ethical questions and uncertain future applications. Perhaps one of the more alarming functionalities enabled by AI advancement is the creation of deepfakes.

Deepfakes undermine our trust in the information we see. Plus, they present plenty of opportunities for mischief and malicious use. It’s hard to see how we could hope to show such functionality in a positive light — or how anyone could ever come to trust it.

But, as with any tool, there are beneficial uses for deepfakes too. Here, we explore some oft-forgotten positive deepfake examples. How can deepfake technology be a force for good?


What is a deepfake?

The term deepfake is a portmanteau of ‘deep learning’ and ‘fake’. Deepfakes are convincing content depicting artificially constructed events.

That content could be a video, photo, or audio recording. Whatever the medium, its content has undergone a verisimilar change via artificial rendering. (Think changes to a face, a voice, or any element of a person’s movements, speech, or actions.)

This most often affects celebrities and predominant public figures. Their likeness is used, but the content isn’t real. They aren’t truly saying those words, or physically enacting those movements.

To be a true deepfake, the creation of this artificial content must involve the use of AI and machine learning functionality. Specifically, deep learning. Some people think of it like photoshop on steroids. (And powered by artificial intelligence.)

This, understandably, has raised a sea of concerns. Aggravating this alarm is the fact that plenty of stories about deepfakes discuss deepfake pornography, fake news and fraud. But where there’s bad, there’s good. Positive deepfake examples also exist.

So, let’s look at some of the beneficial applications of the technology.


Educational deepfake examples

Deepfake technology holds positive potential for education. It could revolutionise our history lessons with interactivity. It could preserve stories and help capture attention. How? With deepfake examples of historical figures.

For instance, in 2018 the Illinois Holocaust Museum and Education Centre created hologrammatic interviews. So, visitors could talk to and interact with Holocaust survivors. They could ask questions and hear their stories. As deepfake technology advances, this kind of virtual history could become achievable on a much wider scale.

Another example comes from CereProc, a company that ‘resurrected’ JFK in voice. This deepfake made it possible to hear the late president deliver the speech he would have delivered, if not for his assassination.

In this way, deepfake technology could help us preserve not just the facts in history books, but the impact historical events had on real people.


Reaching worldwide audiences

Positive deepfake examples also show how the technology can make language barriers (and bad dubbing) a thing of the past.

Take, for instance, the David Beckham malaria announcement. By using AI technology, David Beckham was shown to speak nine different languages in order to share a message for the Malaria Must Die campaign.

Because deepfakes can replicate voices and change videos, it can allow for translated films that use the original actors. The voices sound like the original ones. And, crucially, the lip movements even match the words spoken.

So, with the help of positive deepfakes, we can better share thoughts, films and other creative works on a worldwide basis. Even those with lower budgets. This stands to improve the diversity of our entertainment and content consumption.

In other words, deepfakes can shatter language barriers, making content more accessible.


Deepfake characters

In fact, there are a lot of positive deepfake examples for use in the entertainment industry. Deepfakes can keep film characters consistent.

For example, consider the times that an actor has passed away. Deepfake technology can fill the role of CGI, recreating the likeness of unavailable past actors. So, the character doesn’t have to pass away with their actor. For example, the recreation of the late Peter Cushing in Star Wars: Rogue One (2017), who passed away in 1994.

Or, consider times when a character needs to be older or younger than their actor. For example, the late Carrie Fisher’s character, Princess Leia. Even though the actress herself was not available, her young likeness was recreated. This also demonstrates another positive use of deepfake technology: ageing and de-ageing.


Positive deepfake examples in the art world

Moving away from film, it’s also possible to find positive deepfake examples for the art world.

AI technology could help us create virtual museums. This would allow access to the world’s masterpieces for people that otherwise might not be able to experience them in person. We could share convincing, deepfake artwork across the world.

Deepfake technology could even allow us to resurrect dead artists. For instance, Salvador Dali at the Salvador Dali Museum in Florida.

Perhaps a more novel use of deepfakes could be to bring art to life. Samsung’s AI research laboratory has allowed the Mona Lisa to move her head, eyes and mouth. So, if you thought she was watching you before…


Deepfake examples in medicine

The technology behind deepfakes can also provide benefits to the healthcare industry.

Specifically, it can provide a boost to data privacy, while helping with the development of new diagnosis and monitoring practices.

By using the technology behind deepfakes, hospitals can create deepfake patients. That is, patient data for testing and experimentation that’s realistic, but doesn’t put real patients at risk. So, researchers can use true-to-life deepfake patients, instead of real patient data.  

From this, there’s room to test new methods of diagnosis and monitoring. Or even train other AI to assist with medical decision making.


Training

In fact, deepfake examples could prove instrumental in all sorts of training environments.

Artificial intelligence requires masses of data to train. Often, there are concerns about where all this data comes from. There’s also the issue of algorithmic bias caused by a lack of diversity in data. With deepfake technology, though, we can create true-to-life, diverse artificial data. And so, the problem of needing more data for training could cease.

Or, consider any training for onboarding team members. For example, customer service training. It’s possible that we could one day have realistic, deepfake virtual humans. This could provide deepfake examples of real customers. That way, you can train your new team members, without throwing them headfirst into real customer interactions (for instance.)


Earning trust: proceed with caution

When it comes to deepfakes, there’s plenty of potential for the technology, good and bad. Naturally, the desire is to see more positive deepfake examples, rather than fraud and negativity.

But this can’t happen until the concerns are answered, the correct cautions are taken, and we find a way to trust the tech.

This means that we need to see when something is real and when it’s AI-generated. We need rules that keep the technology helpful, and practices that prevent it from causing harm.


Useful links

ELI5: what is deep learning?

A history of automation: The rise of robots and AI

AI acceptance and the man on the Clapham Omnibus