Get Started with Salesforce AI

Question 1

What are the 3 current waves of the AI revolution?

Answer 1

Predictive AI, Generative AI and Autonomous and Agent AI

Question 2

What is the name of the platform Salesforce AI is built on?

Answer 2

Einstein 1 - provides a unified source of truth

Question 3

What is Predictive AI

Answer 3

Uses large amount of data and machine learning to identify patterns and predict behavior

Question 4

What is Generative AI?

Answer 4

Machine learning plus summarizing information, answering questions, generate images and convert text to speech. Built on NLP. Also known as Einstein gateway, the gateway. The gateway exposes normalized APIs to interact with foundation models and services provided by different vendors, internally and from the partner ecosystem.

Question 5

What is NLP

Answer 5

Natural Language Processing - A branch of AI that uses machine learning to understand language as written by people. Large language models are one of many approaches to NLP.

Question 6

What is Autonomous and Agent AI?

Answer 6

Requires no direct human input. This has not been fully developed yet but is in the works.

Question 7

What protects customer data?

Answer 7

Einstein Trust Layer that sits on the Hyperforce model

Question 8

What LLM?

Answer 8

Large Language Model - A language model consisting of a neural network with many parameters trained on large quantities of text.

Question 9

What is Einstein 1 Studio?

Answer 9

Suite of 3 builders for creating AI

Question 10

What 3 builders are included in Einstein 1 studio and what do they do?

Answer 10

1. Copilot Builder - Conversational AI assistant for CRM - custom actions using AI prompts, flows, APEX, and MuleSoft APIs.
2. Prompt Builder - Custom no code prompts
3. Model Builder - Build or bring you rown model to Salesforce and create solutions that are deployable across CRM systems.

Question 11

Artificial intelligence (AI)

Answer 11

A branch of computer science in which computer systems use data to draw inferences, perform tasks, and solve problems with human-like reasoning.

Question 12

Bias

Answer 12

Systematic and repeatable errors in a computer system that create unfair outcomes, in ways different from the intended function of the system, due to inaccurate assumptions in the machine learning process.

Question 13

Corpus

Answer 13

A large collection of textual datasets used to train an LLM.

Question 14

Domain adaptation

Answer 14

The process through which organization-specific knowledge is added into the prompt and the foundation model.

Question 15

Fine-tuning

Answer 15

The process of adapting a pre-trained language model for a specific task by training it on a smaller, task-specific dataset.

Question 16

Generative Pre-Trained Transformer (GPT)

Answer 16

A family of language models that’s trained on a large body of text data so that they can generate human-like text.

Question 17

Grounding

Answer 17

The process through which domain-specific knowledge and customer information is added to the prompt to give the model the context it needs to respond more accurately.

Question 18

Hallucination

Answer 18

A type of output where the model generates semantically correct text that is factually incorrect or makes little to no sense, given the context.

Question 19

Human in the loop (HITL)

Answer 19

A model that requires human interaction.

Question 20

Inference

Answer 20

The process of requesting a model to generate content.

Question 21

Inference pipelines

Answer 21

A sequence of reusable generative steps stitched together to accomplish a generation task. This includes resolving prompt instructions, passing it to an LLM, moderating the results, and sending results back to the user.

Question 22

Machine learning

Answer 22

A subfield of AI specializing in computer systems that are designed to learn, adapt, and improve based on feedback and inferences from data, rather than explicit instruction.

Question 23

Model cards

Answer 23

Documents details about the model’s performance as well as inputs, outputs, training method, conditions under which the model works best, and ethical considerations in use.

Question 24

Parameter size

Answer 24

The number of parameters the model uses to process and generate data.

Question 25

Prompt chaining

Answer 25

The method of breaking up complex tasks into several intermediate steps and then tying it back together so that the AI generates a more concrete, customized, and better result.

Question 26

Prompt design

Answer 26

Prompt design is the process of creating prompts that improve the quality and accuracy of the model’s responses. Many models expect a certain prompt structure, so it’s important to test and iterate them on the model you’re using. After you understand what structure works best for a model, you can optimize your prompts for the given use case.

Question 27

Prompt engineering

Answer 27

An emerging discipline within AI focused on maximizing the performance and reliability of models by crafting prompts in a systematic and scientifically rigorous way.

Question 28

Prompt injection

Answer 28

A method used to control or manipulate the model’s output by giving it certain prompts. With this method, users and third parties attempt to get around restrictions and perform tasks that the model wasn’t designed for.

Question 29

Prompt instructions

Answer 29

Prompt instructions are natural language instructions entered into a prompt template. Your user just has to send an instruction to Einstein. Instructions have a verb-noun structure and a task for the LLM, such as “Write a description no longer than 500 characters.” Your user’s instructions are added to the app’s prompt template, and then relevant CRM data replaces the template’s placeholders. The prompt template is now a grounded prompt and is sent to the LLM.

Question 30

Prompt management

Answer 30

The suite of tools used to effectively build, manage, package, and share prompts.

Question 31

Prompt template

Answer 31

A string with placeholders that are replaced with business data values to generate a final text instruction that is sent to the LLM.

Question 32

Retrieval-augmented generation (RAG)

Answer 32

Retrieval-augmented generation (RAG)

Question 33

Semantic retrieval

Answer 33

A scenario that allows an LLM to use similar and relevant historical business data that exists in a customer’s CRM data.

Question 34

System cards

Answer 34

An expansion of the concepts within, and application of, model cards to address the complexity of an overall AI system, which may integrate multiple models. For LLM-based systems this includes the core components of model cards (for example, performance, use cases, ethical considerations) plus how the system operates, what models it uses, as well as how content is chosen, generated, and delivered.

Question 35

Temperature

Answer 35

A parameter that controls how predictable and varied the model’s outputs are. A model with a high temperature generates random and diverse responses. A model with a low temperature generates focused and more consistent responses.

Question 36

Toxicity

Answer 36

A term describing many types of discourse, including but not limited to offensive, unreasonable, disrespectful, unpleasant, harmful, abusive, or hateful language.

Question 37

Trusted AI

Answer 37

Guidelines created by Salesforce that are focused on the responsible development and implementation of AI.