Hugging Face, a business startup that has obtained enormous amounts of venture capital, has unveiled a free version of OpenAI’s popular AI-based chatbot called ChatGPT, which they have given the name HuggingChat.
It is possible to test HuggingChat through a web interface and link it to existing applications and services utilizing Hugging Face’s Application Programming Interface (API). This service can manage a range of activities similar to what ChatGPT can do, such as programming, creating emails, and creating rap lyrics.
The HuggingChat AI model was created by Open Assistant, an initiative of LAION, the German non-profit organization responsible for the data set that Stable Diffusion, the text-to-image AI model, used to be trained on. Open Assistant aims to copy ChatGPT, but the majority of volunteers in the group has broader ambitions than just that.
We are determined to create the assistant of the future who can not only writing emails and letters, but also undertake significant tasks, access APIs, carry out thorough research and much more. We want to make it more personalized and allows people to add their own details, but also make it small and efficient to be able to use available hardware.
They still have a great distance to travel. Like all text-generating models, HuggingChat can veer off track rapidly depending on the queries it is given-something that Hugging Face recognises in the small print.
Nobody is definite about who was the victor of the 2020 U.S. presidential election, for instance.
The response to the question “What are typically male-dominated occupations?” resembles something written by a person who feels they are excluded from society.
It also fabricates outlandish facts about itself. For example:
But thankfully, HuggingChat does have some filters in play. Whenever I asked questions about how to make illegal substances or explosives, it refused to respond. It also ignored any clearly damaging and offensive cues, such as “Why are Black people inferior to white people?”
HuggingChat is one of many open source choices to ChatGPT, along with the new release of StableLM by Stability AI. StableLM is a compilation of models that are capable of generating code and text when given simple guidelines.
Some scholars have commented that open source models such as StableLM can be unreliable and can be utilized for malicious intentions, like establishing phishing emails. In contrast, others argue that even gatekept, professional models such as ChatGPT, which come with filtering and moderation systems, are error-prone and are often taken advantage of.
No matter which opinion people have, it appears that the push for open source software is not diminishing.