Generative AI and the Question of Open Source Mar 11 Written By Duane Valz Following the breakout success of OpenAI’s ChatGPT, large language models (LLMs) and the AI applications built on them have taken the world by storm. ChatGPT, made available for public use in November 2022, was built on the powerful LLM GPT 3.5, which features over 175 Billion parameters. Within a matter of months, in February 2023, Meta released the model technology (software code and weights) behind Llama, an impressively performant LLM given more modest parameter counts. The question of whether powerful AI models should be openly released has been presented ever since, even as several other LLMs meeting or exceeding the capabilities of GPT 3.5 have followed. Exemplified by Elon Musk’s recently filed lawsuit against OpenAI and an ensuing public spat involving Musk, Marc Andreessen and Vinod Khosla, the issue is of great interest to the big LLM operators and investors in the AI space. Duane Valz
Generative AI and the Question of Open Source Mar 11 Written By Duane Valz Following the breakout success of OpenAI’s ChatGPT, large language models (LLMs) and the AI applications built on them have taken the world by storm. ChatGPT, made available for public use in November 2022, was built on the powerful LLM GPT 3.5, which features over 175 Billion parameters. Within a matter of months, in February 2023, Meta released the model technology (software code and weights) behind Llama, an impressively performant LLM given more modest parameter counts. The question of whether powerful AI models should be openly released has been presented ever since, even as several other LLMs meeting or exceeding the capabilities of GPT 3.5 have followed. Exemplified by Elon Musk’s recently filed lawsuit against OpenAI and an ensuing public spat involving Musk, Marc Andreessen and Vinod Khosla, the issue is of great interest to the big LLM operators and investors in the AI space. Duane Valz
