OpenAI Broke Hide And Seek
Ғылым және технология
OpenAI Broke Hide And Seek
For many years, AI researchers have been trying to create intelligent machines by studying and replicating the structure and functionality of the human brain. The latest research from OpenAI intended to look into the possibility of AI agents learning sophisticated and real world relevant techniques without any suggestions from the researchers. They simply put agents in a simple game of hide and seek, where they used objects in unexpected ways to achieve their goals. So in this video, we will take a look at how OpenAI tried to train AI agents to play hide and seek and why they were completely shocked? Make sure to watch the video until the end to know how these types of self-taught AI agents could be helpful in the real world as well. Gaming has always been used to evaluate the advancement of AI. This is why researchers are trying their best to unite AI and games to help machines learn from gaming data.
In today's video we look at OpenAI Broke Hide And Seek...Keep watching to see multi agent ai hide and seek Recently open ai plays hide and seek. The results left many researchers speechless. Many have seen ai learns to play video games but this experiment broke hide and seek. AI learns to play hide and seek has allowed insight into emergent behavior of AI. Open ai hide and seek may be the answer to many questions about sentient ai.
Subscribe for the latest news on mechatronics, tech, and futurology. Inspired by Tech Space, Digital Engine, and TechZone.
Inspired by OpenAI Plays Hide and Seek…and Breaks The Game! 🤖
Inspired by Multi-Agent Hide and Seek
Inspired by AI Plays Hide and Seek...Surprises Creators (Bonus Scenes)
Inspired by AI Destroys Hide-and-Seek: Discovers Cooperation & Coordination
Inspired by Open AI Hide & Seek Revolutionizes Reinforcement Learning
Also check out: • Stunning Robotic Hands...
On Blue Light Technology we will go through mechatronics engineering, metaverse, and deep tech. Stay tuned for the latest ai news, technology news, and tesla news.
Subscribe - bit.ly/SubscribeToBLT
Subscribe - bit.ly/SubscribeToBLT
#ai
#google
#openai
Пікірлер: 57
Could you beat Open AI in a game of hide and seek?
@zander3523
Жыл бұрын
Yes
@alvonanimations7531
Жыл бұрын
Yea I’m built diff
@efrengonzalez5638
Жыл бұрын
Probably not
@keatonwallace1828
Жыл бұрын
Yeah ill just lock walls around the seekers
I feel bad for you, I've seen your content stolen on ticktock probably over 15 times before anyone gave credit to you
@bluelighttechnology
Жыл бұрын
that's alright, we appreciate you watching our content!
@DullFool
Жыл бұрын
@@bluelighttechnology that’s actually how I found you video, you should probably do something about that
@Andrew-km5ue
Жыл бұрын
It’s not only their content, there have been other people who have commentated this who have been a lot more popular.
@hi7535
Жыл бұрын
Tiktokers be having the audacity to take a full KZread video and put it into parts. I never watch more than one part and if the video interests me enough I just go find it on KZread so they don't get more views from me.
@DullFool
Жыл бұрын
@@hi7535 same
Here is the best strategy: don't hide runners behind the objects, just lock seekers with objects, so they can't jump or move.
@LyrLyrPantsOnFyr
Жыл бұрын
I’m pretty sure u can only lock from the inside but idk
Wai wait wait so the ai learned to exploit the environment given to them the same way we as humans find shortcuts to better solve problems its incredible yes but in a video game they found a glitch so they win but irl this THIS IS THE REALLY IMPORTANT STUFF
@bluelighttechnology
Жыл бұрын
Exactly. What people might not do because of ethics an AI might still make that decision because it’s more efficient. However it could be unethical.
@althron4249
Жыл бұрын
@@bluelighttechnology And if that unethical thing becomes a choice between solve world hunger by a means where you dont exploit or abuse something the ai might respond with well an unethical answer this is dramatically different but thinking about it is a bit scary but its good because for small tasks it can find shortcuts
"Hide and seek" looks a lot like urban combat
@bluelighttechnology
Жыл бұрын
😅
@erikperik1000
Жыл бұрын
@@bluelighttechnology elaborate
@bluelighttechnology
Жыл бұрын
@@erikperik1000 hopefully that's not why they made this program, to teach it urban combat
Wait til we are training a model to stack boxes more efficiently and it creates portal technology
the little guys are so cute
@bluelighttechnology
Жыл бұрын
Till it becomes conscious and takes over the world 😳
4:17 "Don't think you are, know you are".
Didn’t 2 minute papers do this like 2 years ago?
@Rofusu0001
Жыл бұрын
3 years my bad
@bluelighttechnology
Жыл бұрын
that is correct, we just decided to make our own version.
@Rofusu0001
Жыл бұрын
@@bluelighttechnology ok 👍
I saw a ad rn for this and it said noob vs pro
@bluelighttechnology
Жыл бұрын
Yes we’ve seen that ad for the game someone made
What if the hiders just locked up the seekers
@bluelighttechnology
Жыл бұрын
From what we've researched, the hiders never locked the seekers. So one can only speculate.
How do you think quantum computers would affect this?
@bluelighttechnology
Жыл бұрын
AI would most likely help quantum computers solve problems faster. Powerful machines but can be slow at completing task.
@FunnyAndCleverHandle
Жыл бұрын
@@bluelighttechnology a quantum computer could run more sims at once, right?
@bluelighttechnology
Жыл бұрын
As of now GPUs are the main hardware used to train AI models and algorithms. One specific chip that has been built for AI training is the Cerebras CS-2 but you can use a regular GPU to train AI. Quantum computers are specialized to use quantum mechanics to solve very complex problems. So AI could be used to find a possible solution faster and then input that solution into the quantum computer to see if it would work. The AI model would basically be making predictions but would not have any actual data to support its findings. To collect the date and test the solution, that is when the quantum computer would be used.
4:56 Dude! WTF was that?
*Am i the only one who wants to play a game like that to boxsurf?*
@roseanncleaves5901
10 ай бұрын
nope
Is there a game about this? I wanna play it
@bluelighttechnology
Жыл бұрын
No this is an experiment done by OpenAI. They used these avatars so people could visualize the AI playing data. Otherwise it would just be raw data that mainly only computer scientists would understand
Isn’t this something that happened like more than 3 years ago? This isn’t new and looks like a copy of another KZread video posted 3 years ago…
@Idk05271
Жыл бұрын
It’s because they are just talking about something OPENAI done.It’s not a game and was only a simulation OPENAI had assess to and made a video about.
Zamn
Anyone here from tiktok
@bluelighttechnology
Жыл бұрын
Welcome!
how can i download the game?
@bluelighttechnology
Жыл бұрын
This was an experiment done by OpenAI, it’s not a video game.
@RonicTheEgg
Жыл бұрын
@@bluelighttechnology ok
game name?
@bluelighttechnology
Жыл бұрын
It’s not a game
I saw your content being stolen
@bluelighttechnology
Жыл бұрын
The internet is still very much the wild west when it comes to content