OpenAI Plays Hide and Seek…and Breaks The Game! 🤖

Channel:Two Minute Papers · Length:6:02 · Published:2019-10-22 · Views:11011800

Watch on YouTube

Summary

Two Minute Papers covers OpenAI's multi-agent hide-and-seek experiment, in which agents learned emergent tool use and found ways to break the game's rules. It summarizes the 'Emergent Tool Use from Multi-Agent Interaction' paper.

Why HN recommends it

HN mentions it as a fun and slightly worrying example of AI systems behaving in unintended ways when optimizing.

OpenAI's hide and seek is one of my favorites — HN, 2021

Machine learning & deep learning

Subtitles (excerpt)

While we look at the exact rules here, I will note that the goal of the project was to pit two AI teams against each other, and hopefully see some interesting emergent behaviors. And, boy, did they do some crazy stuff. The coolest part is that the two teams compete against each other, and whenever one team discovers a new strategy, the other one has to adapt. Kind of like an arms race situation, and it also resembles generative adversarial network a little. And the results are magnificent, amusing, weird - you’ll see in a moment. These agents learn from previous experiences, and to the surprise of no one, for the first few million rounds, we start out with…pandemonium. Everyone just running around aimlessly. Without proper strategy and semi-random movements, the seekers are favored and hence win the majority of the games. Nothing to see here. Then, over time, the hiders learned to lock out the seekers by blocking the doors off with these boxes and started winning consistently. I think the coolest part about this is that the map was deliberately designed by the OpenAI scientists in a way that the hiders can only succeed through collaboration. They cannot win alone and hence, they are forced to learn to work together.

On Hacker News

Two minute papers has many fun examples of AIs with unexpected behavior, OpenAI's hide and seek is one of my favorites: [VIDEO] — SatvikBeri, 2021-02
"Two Minute Papers" [VIDEO] GTC 2020 Part 2 was pretty good [LINK] Mostly I'm in a rut as well — sendbitcoins, 2020-05
One of the problems we have to be conscious about with self supervised learning is that the already existing problem of transparency and understandability gets worsened by orders of magnitude. If current billion parameter models are opaque, the result of SSL is, or will be, the equivalent of a black hole. How do you understand what the model has learned? Why has it learned that way? What things c… — montenegrohugo, 2021-03

Related videos