Gallery

Enablement Team

Enablement

Enablement Document

Forte Case Study

Process Manager

Enablement Upskilling Plan

AI Learning

Project Assessment

Explore

AI Learning

Team Insight

AI Use Case for Teams Task

Insight Techniques

When writing a prompt provide context allows

Increases Accuracy

Simplifies Chat

Personalize use case

It will still require review but less work do be done

Using Delimiters

Improved Interpretation

Reusable prompt

Structured output

For words use

For code use

Detailed Instruction

What we want

What format we want it to be

Personas

Inform chatgpt who we want it to be to perform a task

Adding persona is a very concise way of receiving a certain type of input

Splitting up complex task

Get ChatGPT to plan

Start new chat

Summarize then process

Custom Instruction

Create presets of instruction to chatgpt to avoid repeatedly provide same information when chatting with GPT.

⁠

You can create a personal how you would like chatgpt to respond

⁠

Multimodality

are AI Models that can process data from various modalities

Text

Audio

Video

Imagery

Multimodal Tasks

Visual question answering

Where you give the model an image along with a questions

Test to image, video and audio generation

Where you give the model an text prompt with an image or simple a text and the model generates a new image.

Socratic Prompting

⁠

https://arxiv.org/abs/2303.08769⁠

⁠

Getting Chatgpt to Ask

This can be beneficial in 2 ways:

We can make Chatgpt to ask ask question and find out if we are missing some information

Good way of providing chatgpt context

DALL-E

Use detailed description

Choose specific styles

Leverage Chatgpt to enhance prompting

Security Concern

Must be approved by the stakesholders

Check terms of use

Configure environment, if necessary

Omit password keys or sensitive data to chatgpt

Prompting

Order

Instruction first before example

Affordance

Inform the system to do an action first prior proceeding

Example adding calcu() to calculate and equation first before explaining the solution

General Learnings

Language model is not thinking its just returning similar patterns that it pick up during the training

GPT 3 - 3.5

Text Input ( Prompt ) → Language Model → Text Output

GPT-4

Text Input and Image ( Prompt ) → Language Model → Text Output

Massive Multitask Language Understanding (MMLU) Benchmark

Multiple choice questions in 57 subjects

Prompt Engineering

Crafting ideal inputs in order to get the most out of large language models

Artificial Intelligence vs Intelligent Automation

Intelligence Automation

Uses cognitive technologies to streamline processes that are prone to errors and would normally be done manually

Assembly Line

Select → Assess → Optimize → Integrate → Automate → Monitor

Component

Artificial Intelligence

Analyze structure and unstructured data to make predictions

Drives: Decision Making

Business Process Management

Optimizes workflow for greater agility and consistency

Dives: Optimization

Robotic Process Automation

Automates manual, complex and repetitive tasks

Drives: Automation

Learnings:

Automating workflows be a combination of AI, Automation and Human Intelligence intuition and expertise.

Key Benefits of Automation

Streamlines Projects

Reduce Costs

Increase accuracy and predictability

Speeds up implementation

Improves scalability

Learnings:

I am not a product of my circumstances. I am a product of my decision - Stephen Covey

Selecting Workflows to Automate

Identify simple yet time-consuming workflows

Choose well-defined processes that can be repeated

Determine if automation will help or improve the workflow

Assess Workflows to Automate

Conduct workflow analysis

List the tasks and the order they’re completed in.

Create diagram to visualize the process

Classify the resource responsible for each task.

Identify if the people working on each task is changing for each task

Collect data on each task

Analyze data to draw conclusions

Identify task for improvement

Optimizing Workflows:

Areas to focus

Rules

Predefined instruction

To optimize the rule

Event

Identify if the event that triggers task should be modified or eliminated

Route

Identify if instruction can be executed that initially done after the other can be done at the same time.

Result

identify if the result matched your expectation

Resource

Human and machines that perform, support or influence work

Identify every available resources

Responsibilities

Roles and what resources will be held accountable for each task’s completion

Redundancies

Task being done more times than necessary

Integrate and automate project workflows

Inventory of tools

Documentation Tool

Communication Tool

Project Management Tool

Data Management Tool

Monitoring Project Workflow

Regularly review workflows and redesign workflows for improvement

Task Mining

Further refine your result to better understand their underlying tasks, identify inefficiencies, and take action.

Prompt Engineering

Should contain:

Instructions

Questions

Input data

Example

It requires:

Domain Understanding

Knowledge of a particular area or field of expertise that enables AI system to process data in a meaningful way,

It allows you to see what good or bad output look like,

It should be a template that can be modified based on specific need.

Stochastic response

The model’s output is always randomly determined

When prompt is use it provided different results.

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.