DEEPFAKE

What Does Deepfake Mean in a Text?

The Quick Answer

Deepfake is an advanced technique for human image and voice synthesis, based on machine learning (ML) and artificial intelligence (AI).

More Detail...

Deepfake technology allows producers to place anyone's likeness and/or voice in an existing image, video or audio clip, and can even be used to create entirely fictional photos from scratch. The term deepfake was first coined in late 2017, by a Reddit user of the same name. Deepfakes are predominantly used in celebrity pornographic videos and revenge porn (referred to as "deepfake revenge porn"), but have also been employed in hoaxes, fake news, financial fraud and entertainment.

In 2019, the chief of a UK subsidiary of a German energy firm paid €200,000 into a Hungarian bank account, after he received a telephone call from someone he believed to be the CEO of the parent company instructing him to do so. However, the caller was actually a fraudster using audio deepfake technology to mimic the CEO's voice.
Also in 2019, YouTube user Eating Things produced a video in which Jon Snow, one of the leading characters in the Game of Thrones (GOT) television series, apologized for season 8 (which was the final season in the series and which was a huge disappointment to many fans).

Deepfakes use auto encoders and decoders to reduce an image to a lower dimensional latent space and reconstruct the image from the latent representation. A Generative Adversarial Network (GAN) can also be utilized, to train the decoder and a discriminator in an adversarial relationship. This improves the quality of the resultant image by eliminating defects. It is difficult to produce a good deepfake on a standard computer, as the technique requires powerful graphics cards and computing power, as well as expertise. However, the technology is constantly evolving and improving, and there are now a number of cell-phone apps available with a limited deepfake capability. (See also shallowfake.)

Image for Deepfake

When I write deepfake, I mean this:

Deepfake is an Advanced Technique for Human Image and Voice Synthesis.

Summary of Key Points

"Advanced Technique for Human Image and Voice Synthesis" is the most common definition for Deepfake in texting. Here are more details on the word type and users:

Deepfake
Definition:	Advanced Technique for Human Image and Voice Synthesis
Type:	Word
Guessability:	2: Quite easy to guess
Typical Users:	Adults, Teenagers, and Under 13s

Examples of Deepfake in Sentences

Here are examples of Deepfake in conversations:

Robbie: I sent the deepfake video you did of me with Hugh Jackman's body to a girl I'm chatting with on Tinder. She wants to meet up tomorrow!
Sal: Ha ha! You're in trouble.

Sam: Did you see the video of Obama saying President Trump is a total and complete dipsh*t?!
Terri: Yeah! Shame he had to be deepfakeD to say what he really thinks!

An Academic Look at Deepfake

As it is formed by merging several words into one, deepfake is classified as a portmanteau word. It is formed by amalgamating the words deep learning and fake. Deepfake can also be considered a neologism. A neologism is a newly coined word or phrase that has recently come into use or has been created to describe a new concept, idea, or phenomenon. Neologisms can be created in any language and can take different forms, such as:

A new meaning for an existing word (e.g., cloud, mood, sick, tweet, viral).
A completely new word (e.g., fomo, selfie, emoji).
A new combination of existing words, or words with prefixes or suffixes (e.g., deepfake, cloutrage, digital detox, situationship, oversharing, antifragile).

Neologisms are often the result of changes in technology, culture, or social norms, and they can be found in fields such as science, technology, politics, and popular culture. Some neologisms become widely used and enter mainstream vocabulary, while others remain obscure or limited to specific communities or contexts.