At the end of this month, I will be giving a presentation on the Open Source AI Definition at an internal event of the organization LLM-jp. This LLM-jp includes most of the research institutions and companies involved in ML/LLM development in Japan, including major corporations such as SoftBank, the parent company of ARM, TOYOTA, NTT, and SONY, as well as many AI Startups. It is probably the best opportunity to explain OSAID to those who most need to understand it in Japan.
I plan to explain the xls sheet managed by @Mer -san, but is it okay to encourage LLM-jp to verify whether the models developed by Japanese companies comply with OSAID 0.0.8? LLM-jp serves as a platform for information sharing in Japanese AI development, and their voluntary verification of various models would be beneficial for OSI as well.
It seems there were over 160 participants that day. Also, it was unusual for a meeting in Japan to receive so many questions. Although I don’t remember all of them, I believe many questions were related to datasets. Towards the end, I was asked whether making a model Open Source would bring business benefits, but I couldn’t provide a satisfactory answer. I only mentioned that it should have similar benefits to general Open Source software, but that in the case of AI models, the chain of contributions seen with something like Linux might be less likely to occur.
In any case, I was very fortunate to be able to introduce OSAID to an audience almost entirely composed of LLM developers affiliated with Japanese companies. Although I didn’t hear any reactions about checking for compliance with OSAID, many of them were already inclined to be open about AI models and research results from the beginning. Hopefully, we will see some positive developments in the future.
Thank you so much @shujisado -san for your awesome presentation! It’s great to have such a wonderful resource in Japanese and being able to engage with the Japanese community!