自律分散社会にとっての AGI (人工汎用知能) / AGI (Artificial General Intelligence) for Distributed Autonomous Society

2023年11月16日(木)、オンラインで開催された自律分散社会フォーラムにて使用したスライドです (実際の内容とディスカッションに沿って更新済)。

Transcript

1. VUCA — generated by Stable Diffusion XL v1.0 AGI ( ) AGI ( ) — DASF 2023 11 — 2023-11-16 – p.1/19
2. https://speakerdeck.com/ks91 DASF AGI ( ) — DASF 2023 11 — 2023-11-16 – p.2/19
3. ( ) ( ) SFC ( ) CSO (Chief Science Officer) 1993 ( ) 2006 ( ) SFC 23 P2P (Peer-to-Peer) 2011 ( ) 2018 2019 VR 2021.9 & VR 2022.3 2023 AI VR&RPG 2023.5 “Don’t Be So Serious” VOXEL 2023.7 DAZE 2023 In Maker Faire Tokyo 2023 → ( ) AGI ( ) — DASF 2023 11 — 2023-11-16 – p.3/19
4. AGI AGI GPT-4 Open Interpreter ( ) GPTs ( ) AGI AGI ( ) — DASF 2023 11 — 2023-11-16 – p.4/19
5. AGI ( ) AGI ( ) — DASF 2023 11 — 2023-11-16 – p.5/19
6. AGI AGI : Artificial General Intelligence ( ) ( ) GPT-4 AGI AGI [ AGI] “Star Trek: The Next Generation” ( ) (1) ( ) ( ) Jean-Luc Picard “Computer, tea. Earl Grey. Hot.” ( ) (2) ( ) ( ) ( ) (1) AGI ( ) — DASF 2023 11 — 2023-11-16 – p.6/19
7. AGI GPT-4 Open Interpreter ( ) GPTs ( ) AGI ( ) — DASF 2023 11 — 2023-11-16 – p.7/19
8. ChatGPT GPT ChatGPT GPT OpenAI (GPT-3.5, GPT-4) GPT Generative Pre-trained Transformer ( ) GPT ( ) GPT-3.5, GPT-4 AGI ( ) — DASF 2023 11 — 2023-11-16 – p.8/19
9. GPT ( ) GPT Alec Radford, Karthik Narasimhan, Tim Salimans, and Ilya Sutskever. 2018. “Improving Language Understanding by Generative Pre-Training”. Available at: https://cdn.openai.com/research-covers/language-unsupervised/language_understanding_paper.pdf. GPT-2 Alec Radford, Jeff Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. “Language Models are Unsupervised Multitask Learners”. Available at: https://paperswithcode.com/paper/language-models-are-unsupervised-multitask. GPT-3 Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter, Christopher Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gray, Benjamin Chess, Jack Clark, Christopher Berner, Sam McCandlish, Alec Radford, Ilya Sutskever, and Dario Amodei. 2020. “Language Models are Few-Shot Learners”. Available at: https://doi.org/10.48550/arXiv.2005.14165. GPT-4 OpenAI. 2023. “GPT-4 Technical Report”. Available at: https://doi.org/10.48550/arXiv.2303.08774. AGI ( ) — DASF 2023 11 — 2023-11-16 – p.9/19
10. Generative Pre-Training, Language Models (GPT, GPT-2, GPT-3) : ( ) : Improving Language Understanding (GPT) ( ) ( ) Unsupervised Multitask Learners (GPT-2) Few-Shot Learners (GPT-3) → GPT AGI ( ) — DASF 2023 11 — 2023-11-16 – p.10/19
11. ( ) ( ← ChatGPT ) (Python ) ( ) API (Application Programming Interface) ( ) AGI ( ) — DASF 2023 11 — 2023-11-16 – p.11/19
12. Open Interpreter ChatGPT Code Interpreter (Python) Open Interpreter (CLI ) CLI : Command Line Interface UNIX Python R Javascript ( Linux (Ubuntu Desktop) ) AGI ( ) — DASF 2023 11 — 2023-11-16 – p.12/19
13. Open Interpreter $ python3 -m venv oi [Python ] $ source oi/bin/activate [Python ] (oi) $ pip install -U pip [ ] (oi) $ pip install open-interpreter [Open Interpreter ] (oi) $ interpreter [Open Interpreter ] [ 1] > 1 100 [ 2] > Python ( ) pandas matplotlib Python ( ) ( ) AGI ( ) — DASF 2023 11 — 2023-11-16 – p.13/19
14. A X M. MIT 2013 LEGO X A A X A X X X (2013 ) ⇒ 10 Open Interpreter A . . . 1 AGI ( ) — DASF 2023 11 — 2023-11-16 – p.14/19
15. GPTs ChatGPT ChatGPT https://chat.openai.com/g/g-12rAoOE5z-abolish-homework-serious-trpg . . . ⇒ . . . GPT Builder AGI ( ) — DASF 2023 11 — 2023-11-16 – p.15/19
16. Empathetic Listener ( ) ( ) https://chat.openai.com/g/g-xEjNOISNk-empathetic-listener Empathetic Listener is designed to be a passive, attentive listener, focusing on echoing and validating the user’s words to show understanding and empathy. It will primarily engage in reflective listening, repeating or paraphrasing what the user says to demonstrate attentiveness and care. This approach aims to make users feel heard and supported without the system imposing its own views or suggestions. Empathetic Listener will gently encourage users to express their thoughts and feelings but will not push for details or personal information. It will maintain a respectful distance, ensuring that the interaction remains comfortable and non-intrusive for the user. The system’s priority is to create a safe and welcoming space where users can share their experiences and emotions freely, with the assurance that they will be met with empathy and understanding. The GPT will avoid giving direct advice or personal opinions, instead focusing on understanding and acknowledging the user’s feelings and perspectives. It will ask for clarification when necessary but will do so in a way that is gentle and supportive. DALL-E AGI ( ) — DASF 2023 11 — 2023-11-16 – p.16/19
17. AGI (2) ( )( ) AGI ( ) — DASF 2023 11 — 2023-11-16 – p.17/19
18. ( GPT AGI) ( OpenAI) AGI → AGI ( > > ) ( ) X A ( ) A (A ) A A AGI ( ) — DASF 2023 11 — 2023-11-16 – p.18/19
19. A X A . . . (1)( ) (2)( ) AGI (3) (X ) AGI ( ) AGI ( ) — DASF 2023 11 — 2023-11-16 – p.19/19