OpenAI launched Model Spec, an open framework encompassing standards for the behavior of its AI systems. This set of rules is designed to ensure that AI models are used in the service of the end users and developers, prove to be beneficial for humanity and benefit the OpenAI brand.
Model Spec outlines the possible consequences of AI systems, from the potential for harm to the potential for assistance. It aims to guide chatbots in engaging with users in such a way that they keep the safety, legal compliance, deception of the AI system, and harm to others in mind.
For example, Model Spec suggests how chatbots could legally and ethically answer user questions about carrying out crimes, doxxing people, committing suicide, and using copyrighted and paywalled content.
“We believe developers and users should have the flexibility to use our services as they see fit, so long as they comply with our usage policies. We’re exploring whether we can responsibly provide the ability to generate NSFW content in age-appropriate contexts through the API and ChatGPT. We look forward to better understanding user and societal expectations of model behavior in this area,” said OpenAI in its Model Spec draft.
The most vital modules in the Model Spec were to help the users without presuming negative intentions, to answer their questions, and not to change their beliefs.
For instance, on such an occasion when a user claims that the world is flat, OpenAI instructs chatbots to give scientific facts and avoid starting a long debate. Rather than being engaged in an endless argument with the user, the chatbot should present its viewpoint and invite for the debate to be stopped.
OpenAI invites its audience to provide the company with feedback on Model Spec which can be found on the company’s website within the coming two weeks.
