# Game AI Development With OpenAI Universe | Two Minute Papers #125

## Метаданные

- **Канал:** Two Minute Papers
- **YouTube:** https://www.youtube.com/watch?v=vaFhLAbPi8w
- **Дата:** 05.02.2017
- **Длительность:** 2:28
- **Просмотры:** 26,061
- **Источник:** https://ekstraktznaniy.ru/video/14716

## Описание

OpenAI Universe + blog post:
https://openai.com/blog/universe/
https://universe.openai.com/

Also, make sure to check out Google DeepMind's lab:
https://github.com/deepmind/lab
https://deepmind.com/blog/open-sourcing-deepmind-lab/

For the record: no, I am not an Edge user. :)

Terrain learning footage credit: http://www.cs.ubc.ca/~van/papers/2016-TOG-deepRL/

WE WOULD LIKE TO THANK OUR GENEROUS PATREON SUPPORTERS WHO MAKE TWO MINUTE PAPERS POSSIBLE:
Sunil Kim, Daniel John Benton, Dave Rushton-Smith.
https://www.patreon.com/TwoMinutePapers

Subscribe if you would like to see more of these! - http://www.youtube.com/subscription_center?add_user=keeroyz

Music: Dat Groove by Audionautix is licensed under a Creative Commons Attribution license (https://creativecommons.org/licenses/by/4.0/)
Artist: http://audionautix.com/

Thumbnail background image credit: https://pixabay.com/photo-821568/
Splash screen/thumbnail design: Felícia Fehér - http://felicia.hu

Károly Zsolnai-Fehér's links:
Face

## Транскрипт

### Segment 1 (00:00 - 02:00) []

Dear Fellow Scholars, this is Two Minute Papers with Károly Zsolnai-Fehér. OpenAI's Gym was a selection of gaming environments for reinforcement learning algorithms. This is a class of techniques that are able to learn and perform an optimal chain of actions in an environment. This environment could be playing video games, navigating a drone around, or teaching digital creatures to walk. In this system, people could create new reinforcement learning programs and decide whose AI is the best. Gym was a ton of fun,... but have a look this one! How is this different from Gym? This new software platform, Universe works not only for reinforcement learning algorithms, but for arbitrary programs. Like a freestyle wrestling competition for AI researchers! The list of games include GTA V, Mirror's Edge, Starcraft 2, Civilization 5, Minecraft, Portal and a lot more this time around. Super exciting! You can download this framework right now and proceed testing. One can also perform different browser tasks, such as booking a plane ticket and other endeavors that require navigating around in a web browser interface. Given the current software architecture for Universe, practically any task where automation makes sense can be included in the future. And, we don't need to make any intrusive changes to the game itself, in fact, we don't even have to have access to the source code. This is huge, especially given that many of these games are proprietary software, so to make this happen, individual deals had taken place between the game development companies and OpenAI. When the company was founded by Elon Musk and Sam Altman, they have picked up so many of the most talented AI researchers around the globe. And I am so happy to see that it really, really shows. There is an excellent blog post describing the details of the system, make sure to have a look! Also, I reckon that our comments section is the absolute best I've seen on YouTube. Feel free to participate or start a discussion, there are always plenty of amazing ideas in the comments section. Thanks for watching and for your generous support, and I'll see you next time!