The Crossroads of Open Source: Openness, Divergence, and the Future of Collaboration in the GenAI Era

Time
2025年8月09日 13:20 ~ 13:50
Speaker
Richard 林旅強
Room
RB105
Collaborative Notes
https://hackmd.io/ry_cucbugl
MandarinElementary
Main Session Track

Abstract

Since November 2023, I have been working at 01.AI, a Chinese AI large model company, where I am responsible for the open-source efforts and developer ecosystem of the Yi series models. During this time, I experienced firsthand the rapid surge of AI innovation and confronted the controversies surrounding “copying and repackaging.” This experience led us to deeply reflect: What does “openness” truly mean in the GenAI era? Is simply releasing model weights equivalent to the open-source ethos of sharing source code? How are they fundamentally different in spirit and practice?

This talk will use my journey as a starting point to revisit the history of open-source software. We will discover that today’s GenAI evolution strikingly mirrors the past: from the dominance of academia and the high cost of computing, to the rise of commercial companies, and the fierce debate between closed and open approaches, much like Microsoft vs. Linux in the past, and OpenAI vs. Llama/DeepSeek today.

However, history does not simply repeat itself. The proliferation of PCs and the internet fueled the open-source boom back then, but will lowering the cost of computing alone usher in a golden age for open-source AI? We will examine a unique challenge in GenAI: the data problem.

Furthermore, why are projects like n8n and Dify.ai adopting licenses such as Fair Code, which are not OSI-approved, and how do these challenge our traditional understanding of open source? Is this a regression in history, or an inevitable evolution?

Finally, I want to address a fundamental question: In an era where AI can write code, what is the future of human-collaboration-based open-source communities? Are we witnessing the dusk of the open-source spirit, or the dawn of a new chapter of co-evolution and human-AI collaboration?

This talk is not only a retrospective and a comparison with the past but also an open invitation to all open-source advocates to engage in a shared reflection: How do we define “openness” in the age of GenAI, and how can we find a path toward co-creation and co-governance?

About the Speaker

Richard 林旅強

Richard 林旅強

Hi 我是 Richard / 強哥。 COSCUP 的老友,一位在開源世界打滾超過 16 年的傳教士與扛轎者。

我的開源之旅,從擔任 COSCUP 志工、以及在中研院「自由軟體鑄造場」(OSSF)工作的日子開始。投身 COSCUP 的機緣,也讓我陰錯陽差踏上了一條西進之路,進入中國的開發者生態圈,曾擔任華為雲 AI 開發者生態負責人、以及零一萬物(01.AI)開源大模型生態負責人,工作重點就是把產業、技術和社群「喬」在一起。

在社群方面,除了是 COSCUP 的老志工,我也曾在 2014 年共同創辦中國第一個加入 OSI 的「開源社」(注意,不是雞排店)。近年來關注即時互動技術和對話式 AI,並加入 RTE 開發者社區擔任轎班成員。

現在,我創辦了 OpenQ Consultation,並擔任 OpenSeed 孵化器的合夥人,主要使命是服務和協助各種企業、AI 新創,甚至剛萌芽的專案,從零到一打造全球化的開源商業模式,讓更多好的技術和產品被世界看見。

我曾翻譯過《開源項目成功之道》、《開發者關係:方法與實踐》,以及合著《藏智於民:開放政府資料的原則與現況》等書籍。平常也愛在臉書上分享各種好笑或有料的內容,歡迎 follow。

(update: July, 2025)