Yohei
@yoheinakajima
🔥1/8 Introducing "🤖 Task-driven Autonomous Agent" An agent that leverages @openai's GPT-4, @pinecone vector search, and @LangChainAI framework to autonomously create and perform tasks based on an objective. "Paper": yoheinakajima.com/task-driven-au… [More 🔽] pic.twitter.com/9BmbuJBAjl
2023-03-29 13:30:44
拡大
Yohei
@yoheinakajima
🚀2/8 The system can complete tasks, generate new tasks based on results, and prioritize tasks in real-time. It demonstrates the potential of AI-powered language models to autonomously perform tasks within various constraints and contexts.
2023-03-29 13:30:45
Yohei
@yoheinakajima
💡3/8 The autonomous agent uses GPT-4 for task completion, Pinecone for efficient search and storage of task-related data, and the LangChain framework to enhance decision-making processes. #GPT4 #Pinecone #LangChain
2023-03-29 13:30:45
Yohei
@yoheinakajima
🎯4/8 The system maintains a task list for managing and prioritizing tasks. It autonomously creates new tasks based on completed results and reprioritizes the task list accordingly, showcasing the adaptability of AI-powered language models. pic.twitter.com/0h5UfSiIrQ
2023-03-29 13:30:45
拡大
Yohei
@yoheinakajima
🔧5/8 To complete tasks, the system uses GPT-4 and LangChain's capabilities, enriching and storing results in Pinecone. This integrated approach allows the AI agent to interact with its environment and perform tasks efficiently.
2023-03-29 13:30:46
Yohei
@yoheinakajima
🧠6/8 The system generates new tasks based on completed task results and prioritizes them using GPT-4. This allows the system to adapt and respond to new information and priorities.
2023-03-29 13:30:46
Yohei
@yoheinakajima
🔮7/8 Future improvements include integrating a security/safety agent, task sequencing and parallel tasks, generating interim milestones, and incorporating real-time priority updates. pic.twitter.com/11h2M4gG3u
2023-03-29 13:30:46
拡大
Yohei
@yoheinakajima
🤝8/8 This new approach paves the way for AI-powered language models to autonomously perform tasks within various constraints and contexts, enabling new applications and opportunities. Big thanks to all involved! #AIResearch #GPT4 #Pinecone #LangChain
2023-03-29 13:30:47
Yohei
@yoheinakajima
📜 APPENDIX 🧵Thread (above) generated by GPT4 based on paper 📄Paper generated by GPT4 based on code 📊Graphs in paper generated by GPT4 based on code 💻Code generated by GPT4 based on prompt *For each, many prompts to adjust initial output pic.twitter.com/paKvDJvwFh
2023-03-29 13:30:47
拡大
Yohei
@yoheinakajima
Backstory 1/5: Honestly, I was just trying to play around w the idea of an "AI founder" after seeing the awesome #HustleGPT movement. That led to this prompt 2 days ago. twitter.com/yoheinakajima/…
2023-03-29 13:30:47
Yohei
@yoheinakajima
@jenny____ai I’m about 30 prompts/iterations in from this starting point, we’ll see where this goes pic.twitter.com/cH8FsoeV5B
2023-03-27 00:39:59
Yohei
@yoheinakajima
Backstory 2/5: About 50 prompts later (dev docs, error codes, etc.), I shared this working prototype. It's amazing that its first task is to create its next task - and it just keeps going. twitter.com/yoheinakajima/…
2023-03-29 13:30:48
Yohei
@yoheinakajima
So this “AI founder” experiment is kind of blowing my mind. I set an objective, and say “your first task is to create your next task”. It then continues to generate and reprioritize its own task list as it executes them one by one. Only hooked up to search now. Kinda scary. twitter.com/yoheinakajima/…
2023-03-27 11:01:43
Yohei
@yoheinakajima
Backstory 3/5: Realized it could be provided any core objective, in this case "make the world a better place". Pretty fascinating to watch but also scary. twitter.com/yoheinakajima/…
2023-03-29 13:30:48
Yohei
@yoheinakajima
Ohhhh, so I can set ANY objective (eg. make the world a better place). Don't think I should open source this one... 🥴 pic.twitter.com/UEWmyGuDqK
2023-03-27 11:09:37
Yohei
@yoheinakajima
Backstory 4/5: Interestingly, when I asked it to generate as many paperclips as possible, it first generated security measures. Which was then picked up by the creator of the Paperclips Apocalypse theory himself. Led to lots of AI safety reading. twitter.com/ESYudkowsky/st…
2023-03-29 13:30:48
Eliezer Yudkowsky
@ESYudkowsky
tfw the AI approaches AGI safety with the straightforwardness of a child and gives it primary attention from step 1, thereby vastly outperforming all the elaborate dances and rationalizations at the actual big AI labs twitter.com/yoheinakajima/…
2023-03-28 09:28:33
Yohei
@yoheinakajima
Backstory 5/5: Sharing the original experiment led to many shared concerns and potential counter measures being shared publicly. Including awareness of what people are likely doing privately. I believe this is a good thing.
2023-03-29 13:30:49
Yohei
@yoheinakajima
Agh, my site has security issues. So found the second best place to post it. linkedin.com/pulse/task-dri…
2023-03-29 13:49:52
Yohei
@yoheinakajima
Let it be free! Open-sourcing “Baby AGI”, a paired down version of the “Task-Driven Autonomous Agent” at 105 lines of code. Three task agents (execution, creation, prioritization) work in harmony… forever. github.com/yoheinakajima/…
2023-04-03 22:28:20
Yohei
@yoheinakajima
This is based on the autonomous agent I shared last week, same set up. I removed all complexity (@Langchain, @Zapier NLA) so people can add those back in the way they see fit. twitter.com/yoheinakajima/…
2023-04-03 22:28:40
納村 聡仁 / Osamura Akinori
@akinoriosamura
wow 105行のコードで「タスク駆動型自律エージェント」のペアダウンバージョンである「Baby AGI」をオープンソース化。 3 つのタスク エージェント (実行、作成、優先順位付け) が調和して機能。 github.com/yoheinakajima/… twitter.com/yoheinakajima/…
2023-04-04 13:08:35