However, Salesforce acknowledges the issues with large language models, including the AI trust gap and the potential for incorrect or incomplete outputs. The company believes that by linking the Einstein Copilot AI tool to data from its own Data Cloud and building its own model, it can reduce some of these issues. Despite this, it is widely accepted that it is currently impossible to completely eliminate hallucinations in large language models. Einstein Copilot is currently in beta, and the Einstein Trust Layer will be generally available across the Einstein platform next month.
Key takeaways:
- Salesforce has announced the introduction of Einstein Copilot, a conversational AI assistant that allows users to ask questions in the context of their work across the platform.
- Einstein Copilot is designed to provide a more efficient and productive way for employees and customers to access generative AI, augmenting their abilities and skills.
- The company is also introducing a 'trust layer', an underlying security, governance and privacy architecture, to give customers more confidence in using Salesforce's generative AI tooling.
- Einstein Copilot is currently in beta with customers, and the Einstein Trust Layer will be generally available across the Einstein platform next month.