IBM to Train AI to Follow Code of Ethics

Tech giant made one more step in artificial intelligence research
19 July 2018   944

Developers from IBM have created an artificial intelligence that dynamically forms norms of ethics, expressed in a set of rules when creating content recommendation algorithms. This is reported by Venture Beat.

The IBM Neural Network is trained in two phases. Initially, the AI ​​receives a set of constrained examples that the system of recommendations should adhere to. By processing the received data, the neural network forms the required norms of ethics. The more a set of examples, the more accurate the limitations will be.

At the second stage the AI ​​works directly with the user, studying his preferences and reaction to various content. Based on their received information, the system forms a list of recommended videos for viewing. In this case, the neural network adheres to the previously obtained ethical restrictions. In the system settings, you can set the priority relationship between the user's interests and ethical standards.

The problem with this approach is the need to compile examples for the AI. For a child, an older generation has to set up the limits. However, in the overwhelming majority of cases, the system remains with the user tete-a-tete. The user can also determine some ethical norms, however, in the same way, he can change them. On this issue, a group of IBM researchers continues to work together with MIT Media Lab. While the developers offer to choose as a mentor, responsible for ethical issues, a friend or a family member.

In 2017, hundreds of researchers and IT experts, including Elon Mask and Stephen Hawking, compiled a list of 23 basic principles that should be followed when developing AI. In the same year, the Institute of Electrical and Electronics Engineers defined ethical standards for artificial intelligence. Now AI learns to form ethics for people. And on June 17, 2018, IBM announced the creation of a neural network capable of arguably arguing with live opponents.

Neural Network to Create Landscapes from Sketches

Nvidia created GauGAN model that uses generative-competitive neural networks to process segmented images and create beautiful landscapes from peoples' sketches
20 March 2019   150

At the GTC 2019 conference, NVIDIA presented a demo version of the GauGAN neural network, which can turn sketchy drawings into photorealistic images.

The GauGAN model, named after the famous artist Paul Gauguin, uses generative-competitive neural networks to process segmented images. The generator creates an image and transfers it to the discriminator trained in real photographs. He in turn pixel-by-pixel tells the generator what to fix and where.

Simply put, the principle of the neural network is similar to the coloring of the coloring, but instead of children's drawings, it produces beautiful landscapes. Its creators emphasize that it does not just glue pieces of images, but generates unique ones, like a real artist.

Among other things, the neural network is able to imitate the styles of various artists and change the times of the day and year in the image. It also generates realistic reflections on water surfaces, such as ponds and rivers.

So far, GauGAN is configured to work with landscapes, but the neural network architecture allows us to train it to create urban images as well. The source text of the report in PDF is available here.

GauGAN can be useful to both architects and city planners, and landscape designers with game developers. An AI that understands what the real world looks like will simplify the implementation of their ideas and help you quickly change them. Soon the neural network will be available on the AI ​​Playground.