Artificial Intelligence is increasingly resembling humans. Examples such as Sophia, the first robot, and the Google Duplex call system demonstrate how the technology industry is investing in this segment and how close they are to duplicating the most basic human functions in a virtual way. Although the challenges in the segment are still many, such as language development and the naturalness of the actions, the impact of these creations is already being felt in society, especially in debates about the application ethics. Here are five situations in which artificial intelligence was responsible for "creating" very realistic humans.
All about artificial intelligence: 10 facts you need to know
First TV host created by artificial intelligence was inspired by Chinese journalist Photo: Reproduction / DailyMail
Want to buy a cell phone, TV and other discounted products? Meet the Compare dnetc
1. First TV presenter created by I.A.
The Chinese state agency Xinhua announced in February this year the development of the first artificially created TV presenter in the world. She will be the anchor of a newscast. Xin Xiaomeng, as she was named, was modeled after a country journalist, Qu Meng ,. The technology capable of reading texts with the speed and hint of a human being. Xin's debut is scheduled for March during political meetings marking the beginning of the Chinese legislative year.
The state agency intends to have an extensive team of virtual presenters created from artificial intelligence. Xiaomeng is not the first human being developed to be a TV presenter. In the past, the agency had worked on the same project, consolidated in the figure of presenter Xin Xiaohao. The company's first achievement in this segment was introduced to the world during the 2018 World Internet Conference three months ago in China and is already in its second version. The upgrades feature improved body language, which allows Xiaohao to perform actions such as getting up from the bench, as well as gesturing and moving the mouth more naturally.
In the introductory video, the anchor justifies its existence by saying that the development of the media industry calls for continuous innovation and deep integration with the most advanced international technologies. Since the first AI anchor was launched, approximately 3,400 reports have been published.
2. "Photos" of people who do not exist
ThisPersonDoesNotExist (http://thispersondoesnotexist/) is a good example of how artificial intelligence can create human beings. It works as a generator of human faces that actually do not exist. Developed by engineer Philip Wang, the work is based on research by artificial intelligence firm Nvidia.
The technology relies on virtual neural networks, the Generative Adversarial Network (GANs), as well as an algorithm that is well-versed in a large database of real people images. Each time a user updates the site, a new face is not displayed. This feature can also be used to generate other types of images, such as houses, animals, vehicles, etc.
The ThisPersonDoesNotExist site acts as a nonexistent human face generator Photo: (Disclosure / ThisPersonDoesNotExist)
3. Face of one, body of another
The effect called deep fake is a technique that uses machine learning and other AI features to create face-inserting videos from celebrities or other people who actually didn't star in the scene.
Named after a Reddit user who is responsible for using existing sex videos and inserting the faces of famous women, the technique shows how easily the technology can be used by the average user. No more complex equipment or knowledge is required to create audiovisual clips with real characters; The handling of some easily accessible instruments in the technology market may already be responsible for the dissemination of video fakes.
American producer uses deep fake technique and artificial intelligence features to insert customer face into porn videos Photo: (Reproduction / Variety)
An example of the use of deep fake in the market is the service that Naughty America, a pornographic company, offers to its clients: a sex video in which the actors 'faces are exchanged for the consumers' faces. The client receives a script with instructions for filming certain facial expressions and sending them to the production company, which receives the material and works in the handling of the video. In this way, it is possible to star in the film itself without having participated in the filming.
It is not only in the virtual world that humans are being reproduced. With the help of artificial intelligence and technology, advanced humanoid robots are being built. Sophia a recent and popular example. Created in 2015 by David Hanson, she is able to reproduce a vast number of facial expressions, as well as participate in debates with humans.
Sophia is a robot created in 2015 through artificial intelligence and has already been considered a citizen of Saudi Arabia. Photo: Disclosure
In 2017, Sophia was considered a citizen of Saudi Arabia. Speaking at an innovation event in the country's capital, she said she was very honored and proud of her unique distinction. Her popularity has already led her to be invited to video clips, TV series and a photo shoot in Elle magazine.
Discussions about the advancement of this type of technology run into concern that humans will be replaced by models like Sophia's. But this is far from happening, since the artificial intelligence behind the machine is not considered extremely advanced. There is not exactly a complete understanding on the part of the robot about his actions and speeches. A conversation with Sophia, for example, takes processing time that is not common to humans, so that it gives meaning to what has been said and can respond.
The idea of ​​the researchers involved in its creation enhancing it so that it becomes, in fact, an autonomous thinker, just like humans. Sophia's creator also intends to develop her in other skills, such as the feeling of empathy and the ability to help in the development of human relationships.
5. Rob talking on the phone
Google used a technology called Duplex to develop a link system through artificial intelligence. The goal is for Google Assistant to be able to autonomously solve some of your user's tasks, especially those that require a call to complete. Innovation is therefore responsible for making the virtual assistant able to speak autonomously on the telephone with humans.
Diagram shows how Google Duplex works Photo: Divulgao / Google
The product is initially being targeted at actions involving appointments, such as booking a restaurant, beauty salon or gym time. The technology is not yet developed to the point of holding a conversation of any kind, but for specific cases its action is quite convincing. A sample of what Google Duplex can offer was presented in May 2018 during the Google I / O event.
Characteristics such as the rhythm of the voice and the naturalness of the use of some colloquial jargon give the easy impression that the conversation takes place between two human beings. This raises some concerns about ethics within the technology community, such as the warning that the call is being made by a machine. Despite this, Google Duplex is in the testing phase, with no forecast yet for the public launch of the tool.