IEEE Spectrum: Meta’s Challenge to OpenAI—Give Away a Massive Language Model”

In a blog post on the Meta AI site, the company’s researchers announced that they’ve created a massive and powerful language AI system and are making it freely available to all researchers in the artificial intelligence community….

Sameer Singh, an associate professor at University of California, Irvine and a research fellow at AI2 who works on language models, praises Meta for releasing the training notes and logbooks, saying that process information may end up being more useful to researchers than the model itself. Singh says he hopes that such openness will become the norm. He also says he supports providing commercial access to at least smaller models, since such access can be useful for understanding models’ practical limitations.

“Disallowing commercial access completely or putting it behind a paywall may be the only way to justify, from a business perspective, why these companies should build and release LLMs in the first place,” Singh says. “I suspect these restrictions have less to do with potential damage than claimed.”

Read the full story at IEEE Spectrum