OREANDA-NEWS  The Russian short vertical video platform Yappy uses artificial intelligence (AI)-based technologies to moderate content and improve the recommendation system in the application. This was announced by Danila Ovcharov, CEO of the platform.

"Our task is to give the authors the mechanisms for creating videos. How can AI help? The first direction is video editing. The second is content generation. The third is moderation and improvement of the recommendation system, which is a kind of social elevator, because algorithms help the author get to the top. Today we have a feed that adjusts to your interests in real time. We use all the capabilities of AI to ensure the convenience of using the platform for authors," he said at the session "The Future of video platforms. Areas of application of AI" of the conference "Digital Industry of industrial Russia" (CIPR-2024).

AI-based technologies are already being used on the national video hosting Rutube, said David Kocharov, general producer of the platform. For example, AI shorts have already appeared there.

"AI-shorts appeared on Rutube - short vertical format videos that we launched together with the digital laboratory D.Lab, which is part of the Gazprom-Media holding. There are various types of animation in these videos, which has diversified the shelf of our content. More complicated, we launched the pilot of the show "IIgry". This is a quiz where artificial intelligence is the leader, as well as inventing contests and tests for bloggers. Personally, I believe that AI is a tool and it should develop independently, and creators should be able to use it and set directions for development," he said.

Rostislav Chernakov, technical director of Rutube, said that at the moment when the video is just uploaded to the national video hosting, the AI is already analyzing it. "It can find 18+ content, shots with weapons, identify copyright infringement. I think when there's too much content with deepfakes, we'll just train our visual transformers, and they'll recognize it. Accordingly, the models give it all out, pass it to the moderator, and suspicious moments are highlighted," he explained.

At the same time, according to Chernakov, in about 5-7 years, the capabilities of the server equipment will become more powerful and it will be much more difficult to recognize deepfakes.

TASS is the general information agency of the CIPR.