AI for every conversation, campaign, and customer

SAFE & RESPONSIBLE AI

• 

  

  Training and generative AI settings

  Define the exact knowledge and historic data used to train your brand-specific models and constrain your generative AI bots from making off-topic decisions. Manage global settings for every model in one place, including generative freedom, knowledge sources, response length, etc.

• 

  

  Supervised learning to mitigate risk

  Rely on human supervisors and approval paths to review any automatically generated knowledge or content before it’s changed or published. Machine-learned enhancements to knowledge or automatically generated content are served up to admins in a clear, simple environment to one-click approve, reject, or modify.

• 

  

  Data protection and governance

  Ensure data remains protected in transit and at rest with industry-standard security certifications and Khoros-wide adherence to principles of least privilege. Customize any re-training schedules or supervision requirements and prevent brand-specific data from flowing back into 3rd party AI services.

• 

  

  Control AI output and prevent hallucinations

  Customize the built-in guard rails in Khoros’ post-processing services to ensure AI generated content is accurate and aligned to brand voice and guidelines. Inspect audit logs and version controls to quickly understand potential changes to output, then revert or remediate.

• 

  

  Bias avoidance

  Leverage generative AI services that achieve the top standards in preventing bias, discrimination, and unethical behavior. For use cases and enterprises with specific contextual bias concerns, like recruiting or finance, Khoros AI can connect to industry-specific LLMs and AI services that have been pre-validated to prevent these specific outcomes.