LLMs for all humanity // how to solve governance problems (AI alignment)?

https://www.linkedin.com/posts/yann-lecun_human-feedback-for-open-source-llms-needs-activity-7119516948505243649-Kgf8

Knowledge management at scale // aka governance problem

How can we organize crowd-sourcing of human feedback for open source LLMs?

wikipedia.org

What are the options?

• Wikipedia style // censorship, no direct motivation (no profit) to participate
• DAO // tokenomics, high scam/hack risks
• Consortium/professional association/foundation (w3c, linux foundation,etc.) // sponsored and controlled by corps/govs/sponsors
• Many competitive startups // new unicorns or small businesses?

https://www.w3.org/

https://www.linuxfoundation.org/

Startups // space for innovations

https://open-assistant.io/

All open source projects begin with people like you. Open source is the belief that if we collaborate we can together gift our knowledge and technology to the world for the benefit of humanity. Are you in? [link]

Open Assistant is a project organized by LAION and individuals around the world interested in bringing this technology to everyone.

The code and models are licensed under the Apache 2.0 license.

We are in the early stages of development, working from established research in applying RLHF to large language models.

https://open-assistant.io/leaderboard

What’s a problem with LLMs? // why do we need this HF

https://www.linkedin.com/posts/xamatriain_interesting-recent-paper-on-llm-hallucinations-activity-7119442371531997184-5t38

https://arxiv.org/pdf/2310.04988.pdf

https://amatriain.net/blog/hallucinations

https://www.linkedin.com/pulse/evaluating-large-language-models-extending-open-llm-sankhadeep-dhar/