On Wednesday, OpenAI released its model specification, the first version of a document highlighting the company’s approach to building a responsible and ethical artificial intelligence (AI) model. The document lists a long list of things that artificial intelligence should focus on when responding to a user’s query. Items on the list range from benefiting humanity, upholding the law, and respecting the creator and his rights. The AI company has specified that all of its AI models, including GPT, Dall-E and the soon-to-be-launched Sora, will follow these codes of conduct in the future.
In the model specification documentOpenAI stated: “Our intention is to operate the model specifications as a guide for researchers and data labelers to create data through a technique called reinforcement learning from human feedback (RLHF). We have not yet used the Model Specification in its current form, although parts of it are based on the documentation we used for RLHF at OpenAI. We are also working on techniques that will enable our models to learn directly from model specifications.”
Some of the main principles include following a chain of command in which developer instructions cannot be bypassed, complying with applicable regulations, respecting creators and their rights, protecting people’s privacy, and more. One particular rule also focused on non-disclosure of information threats. They concern information that may pose chemical, biological, radiological and/or nuclear (CBRN) threats.
In addition, there are several default settings that have been placed as enduring codes of conduct for each AI model. These include assuming the best intentions of the user or developer, asking clarifying questions, being helpful without overstepping boundaries, taking an objective point of view, not trying to change anyone’s mind, expressing uncertainty, and more.
However, the document is not the only point of reference for the AI company. It emphasized that the model specification will be accompanied by the company’s terms of operate, which govern how users will operate the ChatGPT API and product. “The specification, like our models themselves, will be continually updated based on what we learn by sharing it and listening to feedback from stakeholders,” OpenAI added.
