Hugging Face Unveils Open-Source ChatGPT Competitor HuggingChat
Hugging Face has released an open-source generative AI chatbot with the on-brand name HuggingChat. The new chatbot offers an array of text composition services, including computer code, and its design suggests it could serve as a rival to OpenAI’s ChatGPT.
HuggingChat is online as a standalone website from Hugging Face. Developers interested in embedding HuggingChat in existing software can use Hugging Face’s API to incorporate the chatbot. While capable of most of the same feats as ChatGPT, HuggingChat makes it clear that it is unique and different as seen in the screenshot below. When asked HuggingCHat explains that it emphasizes personalization and creating a “unique and authentic connection between users and their avatar friends.” Relatedly, those avatars can be either characters already created for the platform or ones customized for individual users. HuggingChat also made a point of highlighting its training on emotions and empathy in an admittedly subtle putdown about not taking a “neutral stance” like ChatGPT. A brief experiment demonstrated that while HuggingCHat has high ambitions, it’s not at ChatGPT’s level, at least not yet. As an open-source piece of software, it could improve rapidly.
Hugging Face turned to the nonprofit LAION and its Open Assistant AI model to build HuggingChat. LAION is also the creator of the training dataset Stability AI used to make Stable Diffusion. Hugging Face is known for BLOOM and other open-source generative AI models. HuggingChat extends that work into a conversational interface. Stability AI is sharing the road again as it released the first set of its new open-source StableLM models for writing text and code just last week. While HuggingChat doesn’t explicitly state that it wants to take ChatGPT’s place, the group behind Open Assistant is explicit in seeing that as just a starting point.
“We are not going to stop at replicating ChatGPT. We want to build the assistant of the future, able to not only write email and cover letters, but do meaningful work, use APIs, dynamically research information, and much more, with the ability to be personalized and extended by anyone,” Open Assistant’s GitHub page explains. “And we want to do this in a way that is open and accessible, which means we must not only build a great assistant, but also make it small and efficient enough to run on consumer hardware.”