Deepfake Technology for Advertising

February 20, 2023

What if I told you that you can have Ranveer Singh endorsing your clothing business and Hrithik Roshan talking about your local restaurants? This is not science fiction, this is real.

Courtesy: Zomato

Last year, Zomato ran a campaign “Mann Kiya, Zomato Kiya” - where the ad was customized based on the locality. For people seeing the ad in Chennai, Hrithik craved Idly Dosa, and for those in Amritsar, he craved chole bhature. And the restaurants that Hrithik mentioned were different for different cities based on high order value and the dish most frequently ordered.

Did they shoot multiple ads for multiple cities? No. They used AI to create multiple versions of the same ad. This technology is called ‘deepfake’ technology. Deepfake technology uses artificial intelligence to generate or digitally alter faces and voices and generate custom messaging.

🧐  Examples of Deepfake Advertising

Let me give you a few more examples of using AI technology for advertising:

  • Pepsi featured Salman Khan interacting with a deep faked version of his younger self from 1994 in a hilarious exchange. 
  • Shah Rukh Khan, in Diwali 2021, starred in a ‘NotACadburyAd’ series where he was seen endorsing the local merchants. Around 105k versions of the ad were created powered by AI and made possible by a company named Rephrase AI. 
  • All mechanics and dealers of Castrol received personalized Diwali greetings delivered by actor Tiger Shroff. 
  • Paperspace, a machine learning, created a likeness of Leonardo Di Caprio for an ad. 

These AI-led advertising campaigns can revolutionize the industry. With lower production costs and virality built-in, it can be an excellent tool for brands. You can cast any celebrity for your ads, adjust their age based on your creative needs and can use multiple stars, which would have been impossible.

An interesting use case is building up in the film industry. A company Flawless AI produced a PG-13 version of a scene in the movie ‘Fall’ and produced accurate dubbing with lip sync. Creating dubbing with lip sync can be a game changer for the Indian film industry obsessed with remakes (that way, they can recreate south movies even faster).

Courtesy: Flawless AI

😕 Problems and Challenges 

You must be wondering why it has not caught on completely. Well, with great technologies, come greater challenges. The Zomato campaign I mentioned earlier ran into serious issues when they took the name of a restaurant which shared its name with a temple and a lot of people were offended.

To understand what I am talking about, here are some of the ‘bad’ deepfake videos that were floating around:

  • There was a real estate ad made with Elon Musk getting kidnapped that went viral. (made without Elon Musk’s consent)
  • Unreal_Keanu is a deepfake avatar of actor Keanu Reeves, who is all over TikTok. (The actor himself is entirely off social media.)
  • Many political deepfakes have been floating around featuring politicians like Donald Trump, Putin and more, spreading a lot of fake news. 

Every technology has its pros and cons. I am bullish on this use case, for it democratises the use of celebrities in ads and makes them accessible even for smaller brands.

It is important to use this technology safely and take proper consent. Proper guidelines and operating procedures are a must for AI-based advertising to scale well.

Did you know?

If you want to increase the purchase intent of your ad viewers, you must make them imagine consuming your product. To achieve this, show movement in your ad through visuals, a slideshow or a video. For an image, show a movement in progress (e.g. a person about to bite into a burger) rather than a static scene (e.g. the burger is on a plate).

Source: Yim, M. Y. C., Kim, Y. K., & Lee, J. (August 2020). How to easily facilitate consumers’ mental simulation through advertising: the effectiveness of self-referencing image dynamics on purchase intention. International Journal of Advertising, 1-25.

Past Newsletters

Consistency in UI Design

February 18, 2023

Data Driven Design

February 6, 2023
320,000+ people have already
signed up, what are you waiting for?
Subscribe Now