Einstein Generative AI & Trust

What is the primary value that Salesforce aims to uphold while using Einstein generative AI?

Which principle focuses on ensuring that model responses are verifiable and can be corroborated?

What should organizations do before sharing LLM-generated responses with external audiences?

How does Salesforce strive to reduce the environmental impact of its AI models?

What challenge does Salesforce specifically address to ensure safety in generative AI responses?

What is a characteristic of generative AI that users must be cautious about?

Which of the following is NOT one of the five principles of trusted generative AI according to Salesforce?

What is crucial to do after receiving LLM-generated content intended for customers?

Einstein Generative AI & Trust

Salesforce prioritizes data security and accurate, safe AI experiences
Agreements with LLM providers (e.g., OpenAI) prevent private data retention
Salesforce's generative AI follows five principles for trusted AI

Five Principles for Trusted Generative AI

Accuracy: Model responses backed by explanations and sources; human review recommended
Safety: Detects and mitigates bias, toxicity, and harmful responses
Transparency: Models respect data provenance, grounded in client data
Empowerment: Tools augment human capabilities, improve efficiency
Sustainability: Focus on optimized, efficient models with reduced carbon footprint

Reviewing Generative AI Responses

Generative AI is a tool, not a replacement for human judgment
Users are responsible for any LLM-generated responses shared with customers
External responses should align with company values, voice, and tone
Focus on accuracy and safety during review
Accuracy: Check for factual accuracy, ensure details are correct; data up-to-date and accurate
Bias and Toxicity: Actively check for bias and harmful language due potential for inaccurate language present in the training data
Responses don't meet standards? Don't use them or re-generate. Some solutions allow editors for changes

This quiz explores the five principles of trusted generative AI, focusing on accuracy, safety, transparency, empowerment, and sustainability. It highlights the importance of human judgment in reviewing AI responses and discusses Salesforce's commitment to data security and ethical AI practices. > https://help.salesforce.com/s/articleView?id=sf.generative_ai_trust_overview.htm&type=5