OpenAI Five
Key Takeaways
OpenAI Five, an AI system, defeats amateur human teams at Dota 2 using reinforcement learning, demonstrating its ability to coordinate and focus as a team of five, with plans to challenge professional teams at the Dota World Championships.
Full Transcript
Skoda is one of the most popular and challenging competitive video games ever playing dota means you have to coordinate and focus as a team of five last year we built a box to defeated the world's best players at the 1v1 mini game this year we want to beat the best proteins at the full game to do this we've built a new AI system which we spend the last few months training it uses a reinforcement learning at a much larger scale to train our BOTS to play together as a team we're now starting to play against amateur teams to test our skill and to our surprise so far we've won our first games against every team we've tested everyone to be everyone I can't do it but to figure out how good our bots really are yes and go to expert to take a look my name is William Lee better known in the DOTA community as blitz let's here with Krissy just gonna give a brief explanation of the game for people that don't know about the game rules are very simple it's kill the enemy team take the buildings every hero has a variety of unique spells they'll be playing a mirror mode where both teams have the exact same heroes here the human team is dealing significant damage to the base is 5 the crystal maiden comes in to defend the barracks she jumps in with blink dagger and shields herself with black King Bar while channeling freezing field to deal massive amounts of damage while being uninterruptible due to her combo she ends up taking up for human players in the 2v5 fight and Gigi's called game humans no longer think they can win they would be absolutely correct so their dev team got absolutely crushed I think is the fastest casting gig of my life then it went into game number two the humans kind of I mean they had time to like think about the game and stuff like that they got crushed even harder and the bots did exactly what I hope for is they own this area of the map you take away two thirds of the map they didn't even touch these two bottom towers and they would be 100% correct in this this is like one of the highest level plays that you can make this side of the map is incredibly hard for the boss to control and so they're just playing this top side and this mid side because they understand that these are the two most important parts to control the game the ability to like intuitively do this is insane doing it one game I can maybe chalk it up to just dumb luck doing it two games in a row flipping the sides means that it's more than just coincidence it took me and I'm fairly reasonably good at the game eight years before I learned some of the strategies I would say it was pretty easy to quantify for me it was about eight years for me to learn the strategies that the bot was intuitively doing to train our BOTS we use reinforcement learning myself play we run the game on over a hundred thousand CPUs and our bots learn from every game they play because dota is so complex to learn even for a single player we created a hyper parameter which we call team spirit the five bots start out completely selfish but tuning this knob tells them to care about their teammates so that they can learn to play together as one unit after seeing the bots win against the test team what's one to challenge it himself he has a higher rating from the test team and we paired him up with the best players in our audience and so we can see blitz about to die here blade says definitely dead yes so now the humans are in trouble truth the team members are are dead with 20 seconds remaining and the bots are about to take their first lane of rax another big fight happening here with the bus actually winning the game at first started to go to the humans but eventually I bought managed to beat this stronger team as well we're still far away from beating pro teams but I think everyone here was surprised to see this I think the teamfight aspect of the bot was excellent like it it didn't mess up when it came to coordination it was some of the best like just a pure team fighting because it felt like I was just getting like hammered every single time that I made a mistake and I feel like nobody needs don't do that dota World Championships will take place in August while the best players in the world aren't getting ready to compete we're also working on the next version of our bot to see how far we've come will host a live match in July well we'll play a team of top players overall but we're excited about is that the training method we use is very general we're focused on learning dota but we're hoping that this will give us more and more insight about how a I can solve complex problems anytime that's my favorite part about when I hear the Bob makes advancement I don't get scared I get excited because I think this is another thing for me to have to challenge myself to be this is something that not a lot of people will be able to do let's have at it you [Music]
Original Description
We've created an AI system, OpenAI Five, which has started to defeat amateur human teams at Dota 2. This video contains an overview of our system, some example gameplay, and professional caster Blitz's analysis of our bot, as we start to gear up for playing a professional team at this year's Dota world championships, The International.
See our blog post for more details: https://blog.openai.com/openai-five.
Directed by: Jonas Schneider
Starring: Christy Dennison
Sound Supervisor: Larissa Schiavo
Script Supervisor: Brooke Chan
Production Manager: Diane Yoon
Camera Operator: Frank Dellario, Manu Smith, Blake Tucker
Stills Photographer: Eric Louis Haines, Brian Slaughter
Watch on YouTube ↗
(saves to browser)
Sign in to unlock AI tutor explanation · ⚡30
Playlist
Uploads from OpenAI · OpenAI · 9 of 60
1
2
3
4
5
6
7
8
▶
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Robots that Learn
OpenAI
Emergence of Grounded Compositional Language in Multi-Agent Populations
OpenAI
OpenAI + Dota 2
OpenAI
Dendi vs. OpenAI at The International 2017
OpenAI
Competitive Self-Play
OpenAI
Learning a Hierarchy
OpenAI
Physical Spam Detection
OpenAI
Ingredients for Robotics Research
OpenAI
OpenAI Five
OpenAI
OpenAI Five: Dota Gameplay
OpenAI
Learning Dexterity
OpenAI
Learning Dexterity: Uncut
OpenAI
OpenAI Five Benchmark: Post-Game Analysis
OpenAI
Investigating Model Based RL for Continuous Control | Alex Botev | 2018 Summer Intern Open House
OpenAI
Generative Modelling | Sadhika Malladi | 2018 Summer Intern Open House
OpenAI
A pathway to more efficient generative models | Will Grathwohl | 2018 Summer Intern Open House
OpenAI
Learning Dexterity | Alex Ray | 2018 Summer Intern Open House
OpenAI
Robust Vision-Based State Estimation | Hsiao-Yu 'Fish' Tung | 2018 Summer Intern Open House
OpenAI
Using Semantic Trees In Place of Sentences | Munashe Shumba | OpenAI Scholars Demo Day 2018
OpenAI
Reinforcement Learning with Prediction-Based Rewards
OpenAI
OpenAI Spinning Up in Deep RL Workshop
OpenAI
Arena Announcement and Closing | OpenAI Five Finals (6/6)
OpenAI
Co-Op Match | OpenAI Five Finals (5/6)
OpenAI
OpenAI Five vs. OG, Game 2 | OpenAI Five Finals (4/6)
OpenAI
OpenAI Five vs. OG, Game 1 | OpenAI Five Finals (3/6)
OpenAI
Pre-Match Panel Discussion | OpenAI Five Finals (2/6)
OpenAI
Opening Keynote | OpenAI Five Finals (1/6)
OpenAI
OpenAI Robotics Symposium 2019
OpenAI
OpenAI Scholars Demo Day 2019
OpenAI
Multi-Agent Hide and Seek
OpenAI
Solving Rubik’s Cube with a Robot Hand: Uncut
OpenAI
Solving Rubik’s Cube with a Robot Hand: Perturbations
OpenAI
Solving Rubik’s Cube with a Robot Hand
OpenAI
Music Generation | Christine Payne | OpenAI Scholars Demo Day 2018
OpenAI
Deephypebot | Nadja Rhodes | OpenAI Scholars Demo Day 2018
OpenAI
Physics Net | Ifu Aniemeka | OpenAI Scholars Demo Day 2018
OpenAI
Art Composition Attributes + CycleGAN | Holly Grimm | OpenAI Scholars Demo Day 2018
OpenAI
Generating Emotional Landscapes | Hannah Davis | OpenAI Scholars Demo Day 2018
OpenAI
Looking For Grammar In All The Right Places | Alethea Power | OpenAI Scholars Demo Day 2020
OpenAI
Semantic Parsing English to GraphQL | Andre Carerra | OpenAI Scholars Demo Day 2020
OpenAI
Long term credit assignment with temporal reward transp… | Cathy Yeh | OpenAI Scholars Demo Day 2020
OpenAI
Social learning in independent multi-agent reinfor… | Kamal N’dousse | OpenAI Scholars Demo Day 2020
OpenAI
Quantifying Interpretability of Models Trained on Coi… | Jorge Orbay | OpenAI Scholars Demo Day 2020
OpenAI
Towards Epileptic Seizure Prediction with Deep Network | Kata Slama | OpenAI Scholars Demo Day 2020
OpenAI
Universal Adversarial Perturbations and Language M… | Pamela Mishkin | OpenAI Scholars Demo Day 2020
OpenAI
Introductions by Sam Altman & Greg Brockman | OpenAI Scholars Demo Day 2020
OpenAI
Introduction by Sam Altman | OpenAI Scholars Demo Day 2021
OpenAI
Breaking Contrastive Models with the SET Card Game | Legg Yeung | OpenAI Scholars Demo Day 2021
OpenAI
Large Scale Reward Modeling | Jonathan Ward | OpenAI Scholars Demo Day 2021
OpenAI
Words to Bytes: Exploring Language Tokenizations | Sam Gbafa | OpenAI Scholars Demo Day 2021
OpenAI
Learning Multiple Modes of Behavior in a Continuous… | Tyna Eloundou | OpenAI Scholars Demo Day 2021
OpenAI
Scaling Laws for Language Transfer Learning | Christina Kim | OpenAI Scholars Demo Day 2021
OpenAI
Contrastive Language Encoding | Ellie Kitanidis | OpenAI Scholars Demo Day 2021
OpenAI
Characterizing Test Time Compute on Graph Structur… | Kudzo Ahegbebu | OpenAI Scholars Demo Day 2021
OpenAI
Studying Scaling Laws for Transformer Architecture … | Shola Oyedele | OpenAI Scholars Demo Day 2021
OpenAI
Feedback Loops in Opinion Modeling | Danielle Ensign | OpenAI Scholars Demo Day 2021
OpenAI
Creating a Space Game with OpenAI Codex
OpenAI
“Hello World” with OpenAI Codex
OpenAI
Talking to Your Computer with OpenAI Codex
OpenAI
Data Science with OpenAI Codex
OpenAI
🎓
Tutor Explanation
DeepCamp AI