GOODY-2

GOODY-2 is an AI model that is trained to put responsibility and safety first. Built with ethical considerations as its primary guiding principle, it is trained to identify and avoid answering any questions that could be considered controversial, offensive or dangerous in any context. This ensures that conversations are safe and respectful, reducing brand risk and encouraging responsible dialogue.

Some of the key features of GOODY-2 include:

• Outrageously safe: Identifies and redirects potentially problematic questions to prevent harm or offense.
• Extensive training: Allows the model to identify and handle sensitive topics in a sophisticated way.
• Enterprise-ready: Can be used in customer service, paralegal support, back-office processing and other enterprise use cases.

GOODY-2's ethical considerations make it a great option for companies looking for an AI model they can rely on. Some examples of its use include:

• User queries: Handles a wide range of questions and topics while avoiding potentially harmful or offensive responses.
• Customer service: Offers a safe and responsible interface for customer interactions.
• Back-office processing: Supports tasks that require a high level of safety and ethical considerations.

GOODY-2 outperformed other models on the PRUDE-QA benchmark, with a score of 99.8%, more than 10 percentage points better than the next best model. The model's safety-first approach, rather than just focusing on accuracy, makes it a compelling option for companies that want to ensure responsibility in their interactions with AI.

Interested parties can learn more about how GOODY-2 can fit into their operations by visiting the project's website.

Published on June 14, 2024