AI Flashcards
What are the Salesforce AI Trusted Principles
Responsibility, Account, Transparency, Empowerment, and Inclusion
Responsible:
Safeguarding data that is trusted upon.
Accountability
Seeking and leveraging feedback for continuous improvement.
Transparency
Developing transparent user experiences and guiding users through machine-driven recommendations.
Empowerment
Promoting economic growth and employment opportunities for employees and customers.
Inclusion
Respecting societal values for all those affected, not just those who created the AI model.
AI Maturity Model
-Ad Hoc
-Organized and Repeatable
-Managed and Sustainable
-Optimized and Innovated
Ad hoc
No proper team or resources; initial stage.
Review and risk assessment take place; informal advocacy.
Organized and Repeatable
-Executives’ buy-in established.
-Ethical principles are established.
-Building a team of diverse experts.
-Company-wide education.
Managed and Sustainable
-Ethical standards integrated from project inception throughout the life cycle.
-Bias mitigation in build or buy decisions.
-Metrics identified to track progress.
Optimized and Innovation
-End-to-end inclusive design practices.
-Ethical features and resolving ethical debt are formal parts of the roadmap and resources.
Demographic Data:
Statistical data collected for populations (age, gender, race)
Baisis in Ai
When AI systems produce systematically prejudiced results due to errors in the machine learning process
Data Bias
Occurs when the organization doesn’t produce broad and general data or inputs previously prejudiced data.
Algothrim Bias
Occurs when algorithm design favors certain outcomes over others.
Personalization
Uses: Tailoring products, services, and content to individual preferences and needs.
Key Principles: Consent, transparency, control, security, and fairness.
Privacy and Data Protection Laws
GDPR (General Data Protection Regulation)-
Comprehensive data protection law since 2018.
Protects EU residents’ info from businesses.
Fines up to €20 million or 4% of business revenue.
CCPA (California Consumer Privacy Act): Regulates how businesses handle personal info of California residents.
Principle of Least Privilege
Giving users and systems the minimal level of access necessary to perform their job functions.
Benefits: Reduces attack surface, maintains data confidentiality.
Impacts of Poor Data Quality
-Biased outcomes.
-Ethical and social repercussions.
-Reputational damage.
-Decreased user adoption.
-Increased operational costs.
-Misguided business decisions.
-Compromised system performance.
Benefits of Good Data Quality
Accurate targeting of customers.
Effective lead scoring and routing.
Better cross-sell and upsell opportunities.
Valuable insights on accounts.
Increased adoption and trust.
Implementating Data Quality
Required Fields: Ensure critical information is captured.
Field Types: Use appropriate data types.
Validation Rules: Enforce data integrity.
Workflow Rules: Automate data processes.
Duplicate Management: Prevent and resolve duplicates.
Page Layouts: Optimize user interface.
Data Quality Dashboard: Monitor data quality metrics.
Data Enrichment Apps: Enhance data with additional information.
Developing a Data Quality Management Plan
Naming Conventions: Rules for records.
Formatting: Standards for dates and currency.
Workflow: Stages of a record’s lifecycle (creation, update, review, archiving).
Quality Standards: Measure age, completeness, accuracy, duplication, usage.
Roles and Ownership: Assign responsibilities for record changes.
Security and Permissions: Determine levels of privacy and data access.
Monitoring Process: Ensure data quality.
Salesforce AI Tools
Lightning Platform
Salesforce AI Tools
GPT
Vision
Prediction Builder
Next Best Action
Language
Recommendation Builder
Ecommerce
Product Recommendation
Commerce Insights
Salesforce AI Tools
Einstein Sales
Lead Scoring
Opportunity Scoring
Forecasting
Account Insights
Activity Capture
Call Summary
Salesforce AI Tools
Einstein Services
Bots
Case Classification
Case Routing
Article Recommendations
Case Recommendations
Salesforce AI Tools
Einstein Marketing
Engagement Scoring
Engagement Frequency
AI
AI is the branch of computer science focused on developing machines capable of performing tasks that typically require human intelligence.
Stages of AI Evolution
Narrow (Weak) AI: Performs a single human cognitive task effectively (e.g., ChatGPT).
General (Strong) AI: Matches the cognitive capabilities of a human being (e.g., “I, Robot”).
Superintelligent AI: Surpasses human intelligence.
Human Cognitive Abilities in AI
Learning
Perception
Reasoning
Language
Understanding
Functional Components of AI Solutions
Computer Vision
Enables machines to interpret and process visual data from the world, automating tasks or enhancing decision-making through digital image analysis.
Functional Components of AI Solutions
Natural Language Processing NLP
The intersection of linguistics and computer science, enabling computers to understand, interpret, and generate human language meaningfully.
Functional Components of AI
Natural language Understanding (NLU)
Converts unstructured data (human language) into a structured format.
Techniques: Syntax analysis (parsing grammatical structure) and semantic analysis (interpreting meaning through context).
Functional Components of AI
Natural language Generation (NLG)
Transforms structured information into human-like language.
Processes: Data structuring, lexicalization, and text realization.
Modeling
Modeling: Algorithms create models to make predictions or decisions based on new data.
Deep Learning
A subset of ML using multiple layers of neural networks to analyze various data input factors.
Machine Learning (ML)
Involves creating algorithms that can modify themselves without human intervention by learning from data.
Training:
Algorithms are trained using large datasets (training data) to learn data functions.
Example: Email filtering.
Neural Network
Algorithms that model and operate like the human brain, interpreting sensory data.
Robotics
The design, operation, and construction of robots to assist or replace human efforts.
Core Components: Sensors, actuators, and control systems.
Example: Automated vacuum cleaner.
Generative AI
Generates new content (text, images, videos, music) by learning from large datasets using models like Generative Adversarial Networks (GANs). It innovates by mimicking existing styles but is resource-intensive and complex.
Predictive AI
Uses statistical algorithms to analyze historical data to make future predictions, identifying patterns and trends. It requires quality historical data and is less demanding compared to generative AI.
