When Ian Goodfellow, a PhD student from Montreal, went to a computer engineering conference in 2014 to describe new technology for machine learning, he coined the term deep learning.
That was when deepfake technology came into being, and it was later picked up by Facebook, where AI engineers from the company create videos that we now call “deepfakes”.
In terms of just timeline, I would say 2016 to 2017 is when almost everyone started to notice this emergent phenomenon happening online.
On social media, in the news, or on a TV show (look for Barack Obama), deepfakes are inserting themselves into our societies.
An algorithm that did not exist before it’s been trained to find patterns and match them when it sees them again.
Deepfakes created from films and video editing tools use face data collected by artificial intelligence facial recognition and face data faked onto a computer-generated model to make their fakes.
What deepfake technology?
After the recent surge in social media, Deepfake was introduced to the public. This technique spread quickly through Hollywood blockbusters, short films and YouTube videos. With the advent of Deepfake, this new phenomenon began to take over popular websites like Reddit, where NSFW deepfake accounts are created and shared often.
Through technology platforms like Artificial intelligence for emotive computing (AIE), AI deepfakes can now create faces with authentic expressions. Various experts ponder on what implications this may have on society in the future.
Other experts have been working to combat these disinformation campaigns carried out by politically motivated groups who use AI deepfake technology during their campaigns by writing algorithms based on machine learning models that can identify artificially inserted faces.
How Does Deepfake Work?
In this article, you will explore how deepfake technology work.
In deep-fake technology, artificial intelligence is used to alter a person’s image to make them resemble someone else substantially speaking. The AI generates the face and chooses a video frame showing the person, skin tone, facial parameters and some other technical specifications as desired.
Researchers in this occupation is trying to create a data set of real faces to reduce artefact errors and get deeper into lifelike digital faces. Another goal is to broaden the limitations of well-known deep digital fabricator face software.
Including other characteristics such as eye or hair colour in the digitalization process can generate realistic versions of people not so easily distinguishable from real pictures (e.g., photos saved on social media).
With artificial intelligence, computers can identify objects from the image from face recognition. The system is composed of an input section for registering the features of a face that could be distorted and, based on the data, generating a new photo-realistic look.
The deepfake process makes use of Machine Learning techniques and deep neural networks.
In any recent project, every AI decision should start by considering its eventual consequences at each stage while following the law of dual use: “what can the technology do but not do well? What can be done with the technology that is contrary to what it was meant for?”
What are the uses of deepfake?
Deepfake is a synthetic media that re-contextualizes how we see reality through an array of clips being spoofed, made into videos and distributed onto social media.
AI tools re-contextualize how we see reality through an array of digital media, such as videos and photos powered by artificial intelligence. Deepfakes can be used as a superimposed disguise or to manipulate audio and video, removing faces or making them disappear altogether in a social setting.
===Deepfakes disinformation and deception===
Pew Research has found that vulnerable individuals exist in society who can be falsely accused of crimes with the use of deepfake technology. New avenues for fake news manipulation to push viewpoints have been introduced and spoken about extensively by politicians such as Ralph Northam about the danger.
