Misalignment Museum: Big money has begin to flow — trying to scare you into believing that AI must be centrally controlled by big tech
“A new museum is exhibiting a post-apocalyptic world where AGI goes rogue. It's called "Misalignment Museum", and located only 8 blocks from OpenAI's office 😅
Featuring paper clip sculptures - a reference to a hypothetical AGI that kills humanity to maximize paper clip production at all cost.“
“A new museum is exhibiting a post-apocalyptic world where AGI goes rogue. It's called "Misalignment Museum", and located only 8 blocks from OpenAI's office 😅
Featuring paper clip sculptures - a reference to a hypothetical AGI that kills humanity to maximize paper clip production at all cost.“
🤬5❤1🔥1
Speeding up RL w/ LLMs Reading Manuals - Read and Reap the Rewards: Learning to Play Atari with the Help of Instruction Manuals
-Finds & summarizes info from Atari manuals
-LLM evaluates agent actions based on ingested manuals
-Improves on games w/ sparse rewards
-Requires 1000x less training frames than previous SoTA on hardest Atari game
Paper
-Finds & summarizes info from Atari manuals
-LLM evaluates agent actions based on ingested manuals
-Improves on games w/ sparse rewards
-Requires 1000x less training frames than previous SoTA on hardest Atari game
Paper
👍2❤1
Asking BingAI to make 2x2s of various bands along weird dimensions (it is coming up with these entirely on its own)
Look up how to do 2x2 matrices. Then create 2x2 matrices for musical acts, using really interesting dimensions for the 2x2s
Look up how to do 2x2 matrices. Then create 2x2 matrices for musical acts, using really interesting dimensions for the 2x2s
👍6
“Some people on my TL are imagining that we can slow or "press pause" on AI... y'know, for safety reasons. Let's be clear on what that would mean.
Short of global anti-AI totalitarianism, all we'll end up doing is redistributing AI's development and benefits away lawful users in the west & toward criminals & gov'ts elsewhere.
Again, we will not stop AI or even slow it. We will only redistribute the advances in it & the benefits that flow from those advances (including military advantages) away from the west & toward the east & global south.”
Article
Short of global anti-AI totalitarianism, all we'll end up doing is redistributing AI's development and benefits away lawful users in the west & toward criminals & gov'ts elsewhere.
Again, we will not stop AI or even slow it. We will only redistribute the advances in it & the benefits that flow from those advances (including military advantages) away from the west & toward the east & global south.”
Article
👏9❤1💯1
PaLM-E: An Embodied Multimodal Language Model
New research published today by Google: PaLM-E 562
The 562 billion parameter "embodied language model" incorporates continuous sensor data directly into language models.
https://palm-e.github.io/
New research published today by Google: PaLM-E 562
The 562 billion parameter "embodied language model" incorporates continuous sensor data directly into language models.
https://palm-e.github.io/
👍4❤1
This media is not supported in your browser
VIEW IN TELEGRAM
PaLM-E
“We show a few example videos showing how PaLM-E can be used to plan and execute long horizon tasks on two different real embodiments. Please note, that all of these results were obtained using the same model trained on all data. In the first video, we execute a long-horizon instruction "bring me the rice chips from the drawer" that includes multiple planning steps as well as incorporating visual feedback from the robot's camera. Finally, show another example on the same robot where the instruction is "bring me a green star". Green star is an object that this robot wasn't directly exposed to.”
“We show a few example videos showing how PaLM-E can be used to plan and execute long horizon tasks on two different real embodiments. Please note, that all of these results were obtained using the same model trained on all data. In the first video, we execute a long-horizon instruction "bring me the rice chips from the drawer" that includes multiple planning steps as well as incorporating visual feedback from the robot's camera. Finally, show another example on the same robot where the instruction is "bring me a green star". Green star is an object that this robot wasn't directly exposed to.”
👍6❤1
🚨 New Upgrade: The real ChatGPT, in all its annoying glory, will now reply to your questions in the group 🚨
To use:
1. Join the group
2. Type "/ask ___” followed by your prompt,
3. Reply to ChatGPT’s previous answers to continue to chat conversations,
4. Join in on others’ conversations simply by replying to their chats
🚨🚨🚨🚨
To use:
1. Join the group
2. Type "/ask ___” followed by your prompt,
3. Reply to ChatGPT’s previous answers to continue to chat conversations,
4. Join in on others’ conversations simply by replying to their chats
🚨🚨🚨🚨
👍17🤣6❤4👏3🤬1😡1