Personal Autonomous Intelligence Computer (PAIC)

Authors: Hanqing Wu, Zile Yang, Jiahuan Zhang, Dongbai Chen, Baoping Hao, Hao Zhang, Hongying Han, Wenlin Fu, Kaicheng Yu

Abstract: In this position paper, we introduce a concept of personal autonomous intelligence computer~(PAIC), to combine the advantage of two mainstream intelligence systems, connectionism and rule-based symbolism. Similar to a traditional von Neumann architecture, our PAIC consists of the input-output, memory, control, processing, and storage Unit. PAIC not only removes the limitation of the fix-length context window from the large language model but also achieves multi-hop reasoning-related precise manipulation of input with a novel pipeline. We also argue that, with the progressive increase of PAIC users on the Internet, the data flow of the current Internet will be reformed to achieve a decentralized local network.

Paper

Webpage Portal

Architecture Design

kOS: An Operating System for PAIC

Native ACT

Coming Soon

System ACT

Coming Soon

Customized ACT

Coming Soon

If you utilize any aspect of our work, please ensure to reference the following paper.

@article{PAIC,
  title={Personal Autonomous Intelligence Computer},
  author={Hanqing Wu, Zile Yang, Jiahuan Zhang, Dongbai Chen, Baoping Hao, Hao Zhang, Hongying Han, Wenlin Fu, Kaicheng Yu},
  url={https://github.com/KMind-Inc/PAIC/blob/main/Main%20Paper/PAIC_Paper.pdf},
  year={2024},
  month={Feb}
}