![]() ![]() ![]() We’re conducting our own research into manifestations of harmful bias and broader issues in fairness and representation, which will help inform our work via improved documentation of existing models as well as various improvements to future models.We’re working closely with users to understand their use cases and develop tools to surface and intervene to mitigate harmful bias.We’ve developed usage guidelines that help developers understand and address potential safety issues.Here are the steps we’re taking to address these issues: As we discuss in the GPT-3 paper and model card, our API models do exhibit biases that will be reflected in generated text. Mitigating negative effects such as harmful bias is a hard, industry-wide issue that is extremely important. We have tens of thousands of applicants for this program already and are currently prioritizing applications focused on fairness and representation research. We’re starting with a very limited number of researchers at this time and already have some results from our academic partners at Middlebury Institute, University of Washington, and Allen Institute for AI. We are also continuing to conduct research into the potential misuses of models served by the API, including with third-party researchers via our academic access program. Constraints that can make generative use cases safer include systems design that keeps a human in the loop, end user access restrictions, post-processing of outputs, content filtration, input/output length limitations, active monitoring, and topicality limitations. Open-ended applications of the API (i.e., ones that enable frictionless generation of large amounts of customizable text via arbitrary prompts) are especially susceptible to misuse. ![]() One key factor we consider in approving uses of the API is the extent to which an application exhibits open-ended versus constrained behavior with regard to the underlying generative capabilities of the system. As we gain more experience operating the API in practice, we will continually refine the categories of use we are able to support, both to broaden the range of applications we can support, and to create finer-grained categories for those we have misuse concerns about. We terminate API access for use cases that are found to cause (or are intended to cause) physical, emotional, or psychological harm to people, including but not limited to harassment, intentional deception, radicalization, astroturfing, or spam, as well as applications that have insufficient guardrails to limit misuse by end users. In production reviews, we evaluate applications across a few axes, asking questions like: Is this a currently supported use case?, How open-ended is the application?, How risky is the application?, How do you plan to address potential misuse?, and Who are the end users of your application?. We have a mandatory production review process before proposed applications can go live. For the API, we’re able to better prevent misuse by limiting access to approved customers and use cases. With GPT-2, one of our key concerns was malicious use of the model (e.g., for disinformation), which is difficult to prevent once a model is open sourced. ![]()
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |