THE VIDEO SHELF · multi-agent reinforcement learning
OpenAI Plays Hide and Seek…and Breaks The Game! 🤖
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 readers recommended it
HN mentions it as a fun and slightly worrying example of AI systems behaving in unintended ways when optimizing.
This is a summary from the source dataset, not an endorsement by Hacker News or the video creator.
Video & recommendation details
- Creator / channel
- Two Minute Papers
- Topic
- multi-agent reinforcement learning
- Format
- explainer
- Level
- general
- HN recommenders in dataset
- 2
- HN mentions in dataset
- 3
- Recommendation years in dataset
- 2020–2021
These counts and years describe the included Hacker News records; they are not YouTube views or the video’s publication date.
From the original HN conversations
Excerpts from the recorded comments. Follow each source for the full comment and surrounding discussion.
"Two Minute Papers" [VIDEO] GTC 2020 Part 2 was pretty good [LINK] Mostly I'm in a rut as well
sendbitcoins on Hacker News · 2020-05
In Ask HN: What interesting stuff are you browsing?Two minute papers has many fun examples of AIs with unexpected behavior, OpenAI's hide and seek is one of my favorites: [VIDEO]
SatvikBeri on Hacker News · 2021-02
In Medical chatbot using OpenAI’s GPT-3 told a fake patient to kill themselves
Summaries, classifications and recommendation records are from the HackerTube source dataset. Read about this library and its limitations.