Researchers from Facebook have created a model to simulate the operation of social networks. Each time the portal introduced an update to its system, it encountered difficulties in ensuring the privacy, safety and security of its users. In order to minimize the risks resulting from this, a fake social network was built to be tested by bots. WES – Web-Enabled Simulation is a new method of building very realistic, large-scale social network simulations. The system uses machine learning to train bots to realistically simulate the behavior of real people on a social media platform. Their main task is to reproduce the experience of using the portal by adding posts or friends. Thanks to this process the Facebook platform is supposed to be safer, more stable and efficient. But that’s not all – bots can also simulate bad behavior of people. Each of the bots is trained on a different type of personality and can interact with each other – some can pretend to be victims and others can be more vulnerable to fraud. This allows researchers to combat abuse on the platform more effectively. More info here and here.
Apple has patented multipath audio stimulation using audio compressors, which uses bone conduction to transmit sound, replacing traditional headphones placed in the ear. Bone conduction works as follows – the user wears a device that transmits vibrations to the cochlea and then converts them into nerve signals that the brain interprets as sounds. The device is located in the back of the head and is held by a headband – link.
Erica, an android created by scientist Hiroshi Ishiguro from Osaka University in Japan is considered to be the first fully autonomous AI robot operating as an actor to play in the film “b”. Erica is playing herself, an AI robot female, b, which can enter the body and mind of any human host. The film follows the efforts of her creators to gain control over her when she becomes self-aware. Some scenes were already filmed in Japan last year. At first Erica did not understand what acting was about, but it was a team of actors who taught her how to imitate the performances of other artists and explained how actors feel in each scene. The robot remembered his lines flawlessly, but it was difficult to distinguish between when she could say something in a soft voice and when in a hard voice, depending on the context. Then they worked on body language and emotions. More information here.
Stanford scientists have created an AI-powered bot that automatically hides the faces of the Black Lives Matter protesters in the pictures. Instead of blurring the faces, the bot automatically covers them with a black fist emoji that has become synonymous with the Black Lives Matter movement. As the images from the protests were widely distributed in social media, the police used this and added the people to the face recognition databases. Thanks to the system created, the right to protest will be secured and no one will feel threatened – link.
Musk predicts that artificial intelligence will superior humans in 5 years. The scientist has long warned people about the dangers of the AI, claiming that this technology may be more dangerous than nuclear weapons. Musk was a few years ahead of the predictions of a famous futurist, Ray Kurzweil, who predicted that it would happen in 2045 – link.
UNESCO launched a worldwide online public consultation on ethics and artificial intelligence. This allows everyone around the world to participate in an international expert group on artificial intelligence. The group is expected to prepare a document with recommendations on the ethical principles of the AI to be submitted to UNESCO member states by November 2021. The recommendation is intended to define common values and principles as well as concrete policy measures related to the ethics of the AI – link.
This Friday (07.08.2020) – the first art festival created by AI is to take place. Edinburgh Fringe is the world’s largest festival of performing arts, but this year’s event has been cancelled due to COVID-19. The solution is to be an AI called ImprovBot. The system analyzed the text descriptions of each performance shown at the festival between 2011 and 2019, and then used this data to generate ideas for new comedies, plays, musicals and cabarets. ImprovBot aims to create 350 show descriptions, which will be published on Twitter every hour from 7 to 31 August – link.
