Movers and SHAKERS
(Note: companies that could be impacted by the content of this article are listed at the base of the story (desktop version). This article uses third-party references to provide a bullish, bearish and balanced point of view; sources listed in the "Balanced" section)
Deepfakes are fabricated clips using artificial intelligence software that can appear to make a person or image speak or do things that had never happened. Deepfake technology normally uses a large quantity of images set, creating a realistic depiction. A Samsung artificial intelligence lab in Russia was able to develop new software that is able to generate a video by inputting a singular image or photo, such as a painting. New facial recognition software and deepfake technology can be used to provide a seemingly endless amount of entertainment but also may pose a threat to public safety.
The Samsung's artificial intelligence lab called its new
deepfake systems "realistic neural talking heads" relating to the
implementation of neural networks and machine learning into realistic media
production mimicking humans. Samsung researchers believe that the breakthroughs
in deepfake technology can be used in a wide array of media ranging from video games to television and film. Researchers stated, "such ability has
practical applications for telepresence, including videoconferencing and
multi-player games, as well as special effects industry." Samsung’s
deepfake technology uses a "meta-learning stage" watching videos of
human facial expressions and applies them to images to create a realistic new
video. Samsung’s deepfake still does not completely provide a full rendition of
the person, maintaining semblance but missing minute details. Siwei Lyu, a
computer science professor at the University at Albany in New York, believes
that this approach to machine learning will be able to quickly adapt and target
new people using less initial data and iterations.
Entertainment: New deepfake technology can be used for a multitude of purposes such as entertainment value. Samsung was able to bring the classic portrait of the Mona Lisa to life. The Mona Lisa is a sole portrait existing as a still image which was able to be animated through new deepfake technology in three different clips. The deepfake technology used by Samsung would be able to provide a new level of harmless entertainment and immersion bringing pictures to life. It would bring about an entirely diverse new genre incorporating everything from memes to movies.
Communication: Deepfake technology helps to broaden communication by enabling new forms of voice generation and representation. Wavenet a generative model for audio can incorporate voices into everyday applications such as Google Assistant. Developing applications, Lyrebird and Modulate, can imitate your voice with a minimum amount of speaking audio allowing personal representation from off-site. Baidu’s technology can completely clone a voice in under 4 seconds. The Dalí Museum in Florida developed a deepfake to create an accurate historical representation of the Spanish surrealist allowing people to actively interact with him. Synthesia, a London-based tech firm, created a realistic rendition of David Beckham to promote a malaria campaign.
Misinformation: Samsung has been able to create deepfake technology and videos by only using a single photo which could be severely abused in order to spread misinformation. Hany Farid, a Dartmouth researcher who specializes in media forensics to root out deepfakes, stated "Following the trend of the past year, this and related techniques require less and less data and are generating more and more sophisticated and compelling content. These results are another step in the evolution of techniques ... leading to the creation of multimedia content that will eventually be indistinguishable from the real thing." The rapid development of new deepfake technology creates a great risk to both officials and the general public, which could lead to fraud and election tampering.
Hard to Detect: Advancements in deepfake technology bring about concerns surrounding mass deception since they constantly become more difficult to discern from reality. A viral video of Nancy Pelosi, US Speaker of the House, was poorly put together using limited deepfake technology but was still convincing enough to deceive watchers and spread quickly on social media. Even the most basic deepfake software and technologies are able to use artificial intelligence and machine learning to produce a forgery or fabrication of a person. Image and video manipulation using computer software has existed for an extended period of time but the development of new deepfake software has made it increasingly to difficult to detect and easier to curate.
A Darker Path: Deepfake technology can be used for much more dark and insidious reasons than just trying to spread misinformation or promote election tampering. Many recent deepfake technologies have been used to take an average person or celebrity’s face and incorporate it into explicit pornographic films or photos. This type of synthetic media can induce severe consequences for the unsuspecting victim involved and leaves many people vulnerable to such cases of defamation. An application called DeepNude was able to render any persons face onto a realistic nude body. Mutale Nkonde, a fellow at the Data & Society Research Institute, stated, “The DeepNude app proves our worst fears about the unique way audiovisual tools can be weaponized.” Deepfake applications such as DeepNude can cause irreversible reputation damage to a person’s career, and their emotional and mental health.
Public: The constant advancement of artificial intelligence technology
allows for the development of new facial recognition software. Researchers are able
to provide new breakthroughs in deepfake technologies allowing for easier and
faster media generation and recognition. These advancements also allow for the
creation of harder to detect fake media and manipulation which can be used to
defame a person or provide misinformation. Artificial intelligence technology
companies and researchers must actively try to deter to the malicious use of
deepfake software in order to prevent harm and protect the rights of the
general public and consumers.
https://medium.com/twentybn/deepfake-the-good-the-bad-and-the-ugly-8b261ecf0f52, Nahua Kang May 16, 2019