Let’s start with a general overview of deepfakes in general, and why they are concerning governments. Then we will move onto some more positive aspects and how they could change elements of corporate and advertising production.
The rising concern of deepfakes
Deepfakes is a catch-all term for artificial production, manipulation or modification of data and by automated means. Specifically, through the use of AI algorithms, they can change an existing piece of contents original meaning, which in turn could be used to mislead the audience. It’s the ability to mislead an audience so convincingly that is causing the greatest concern amongst governments because it can be used for nefarious reasons.
If the intent of the creator of the deepfake clip is to damage a person’s reputation, say by manipulating a clip to show them cheating on their partner (an obvious choice), or by changing a filmed interaction to show them being racist, then this technology can be very damaging indeed to someone’s career and personal life. The current issue in these scenarios is that there are currently no laws in place to manage this content and the public are for the most part generally unable to determine if a deepfake clip is real or not. This in turn creates two further major issues for governments and law enforcements across the world.
Firstly, until now in a criminal courtroom, any video evidence is seen as disreputable – video footage does not lie, right? Well, now it can. Secondly, and perhaps more worryingly is what is being called ‘Liar’s dividend’ whereby genuine footage of controversial content can be dismissed by the subject as deepfake, even though it is true.
With the technology developing at such a fast pace, governments need to act quickly. With all these things considered we could soon be entering into a time when no one knows what is real or what is fake.
To find out more about the potential impact of deepfakes on society see the AI think tank Future Advocacy – https://futureadvocacy.com/deepfakes/
For some context here’s a deepfake BuzzFeed created in 2017 with Jordan Peele.
Here’s a more recent deepfake of Tom Cruise, this shows us how more believable the technology is becoming
Deepfakes could be the hot new tool for corporate and advertising content production
Now let us set aside all the concerning elements for governments above, and let’s look at some of positive aspects of this technology for businesses and advertisers.
Internal corporate training videos and materials are an important aspect for large corporate companies to communicate with their thousands of staff. Most large corporations have employees who speak different languages and are spread across the globe. Currently these companies would create these training videos in one language, then either subtitle the content for other regions, or indeed adapt the content with bespoke aspects for filming and content. This approach can be expensive and very time consuming.
With the synthetic AI technology behind deepfakes, the content could now be created in the main language, then the voice and mouth movements be adapted to the local language. This will not only create more engaging content for the adapted languages but could also reduce the costs of localisation significantly. Advertising giant WPP did just this, they used a synthetic presenter who spoke the recipient’s language and even addressed them by name. This personalisation aspect is another element that will likely become more prevalent across all aspects of moving image outputs in the future.
Therefore, the technology is not only developing in terms of manipulating existing real footage, but companies using AI technology are also now developing capabilities to create synthetic people – so creating people that look real but have never existed. Get your head around that.
A startup company called Rosebud AI is specialising in creating these synthetic models. This could change the control advertisers have over their content. Let’s take a hypothetical example. Say you are a clothing brand; you could select all the visuals aspects of your models appearance and put them in your clothes to create the most visually appealing image for your target audience. You could then easily adapt the model and create many variations of the image with different model ethnicities, ages etc without the need for expensive reshooting. It would also allow these companies to create many variations of content relatively quickly. So, this could potentially be slightly bad news for fashion models and photographers in the long run. The rights of the image also must be determined – could a company then own a synthetic model’s face and form? This seems very possible.
As the above example’s technologies develop and become ever more realistic and easier to adapt and implement there is a high possibility that advertising companies will increasingly use the technology to create content. We are perhaps not too far away from personalised adverts on our television screens, or actors speaking another language seamlessly for a local audience, or personalised audio/visual billboards, flight safety videos or social media content. The possibilities here are quite staggering to think about. The next question that follows then would be ‘will this technology create an enhanced experience for the audience?’ This is where marketeers need to think about what adds value and what does not. It is exciting times for the advertising industry indeed.
You may also like: 7 Ways to Grow Your Business Using Artificial Intelligence
About the Author
Miles Paulley, Producer – Hocus Pocus Studio