Hugging Face, a well-funded AI startup, recently launched HuggingChat, an open source counterpart to OpenAI's popular AI-powered chatbot, ChatGPT. Accessible via a web interface and integratable into existing apps through Hugging Face's API, HuggingChat is capable of executing tasks similar to ChatGPT, such as drafting emails, composing rap lyrics, and writing code. Open Assistant, a project organized by the German nonprofit LAION, developed the AI model behind HuggingChat. LAION was instrumental in producing the dataset used to train Stable Diffusion, a text-to-image AI model. While Open Assistant's initial goal was to mimic ChatGPT, the primarily volunteer-based group has set its sights on a wider range of objectives.
According to Open Assistant's GitHub page, their aim is to create a highly customizable, extensible, and efficient AI assistant that can carry out meaningful tasks, access APIs, and dynamically search for information on consumer hardware—all while remaining open and accessible to users. Nonetheless, HuggingChat has its limitations, as its performance can rapidly deteriorate depending on the questions posed. Hugging Face is aware of this issue and has acknowledged it in the disclaimer. Furthermore, HuggingChat is not entirely devoid of safeguards, as it will not respond to inquiries about creating dangerous or illegal items such as explosives or drugs. HuggingChat joins a growing list of open source alternatives to ChatGPT, including Stability AI's recent release of StableLM—a suite of models that can generate text and code based on simple instructions.
The release of open source models like StableLM has sparked debates among researchers. While some critics argue that these models can be misused for malicious purposes, like crafting phishing emails, others point to inadequacies and vulnerabilities in commercial, gatekept models like ChatGPT, despite their filters and moderation systems. This ongoing debate highlights that the open source movement continues to gain traction within the tech industry. No-code platforms like AppMaster are thriving, providing accessible and efficient app development options for users. With such a growing community of open source options, the future seems bright for technological advancement and accessibility.