3/19/2024

GENERATIVE AI IN Module 2
MODERN GETTING TO KNOW THE
TOOLS
MARKETING

Module Summary

1
 3/19/2024

Homework 1

• How will generative AI change marketing in the next ten years?

• Based on what you learned today, write an opinion piece (100 to 300 words) on
the above topic.
• Use your favourite AI tool to fine-tune your writing, but retain your own original
ideas.
• Submit your work using the submission link for Homework 1 in your LMS.
• This exercise carries 10% of your final evaluation.

Module Outline
• What are Large Language Models?
• Three types of LLMs
• What is a Multi-Modal LLM
• Essential Generative AI Tools Kit for Marketers
• Chat GPT
• DALLE-E-3
• Google Gemini
• CLAUDE 3
• Miscrosoft Copilot
• Grammarly PRO

2
 3/19/2024

What is a Large Language Model?

What is a Large Language Model?

Large language models (LLMs) are an advanced form of artificial intelligence (AI)
algorithm designed to comprehend, summarize, generate, and predict new
content. These models utilize deep learning techniques, and leverage massive data
sets to achieve their capabilities.

3
 3/19/2024

How LLMs Work

• LLMs excel at generating text that closely resembles human language patterns
and expressions.
• They achieve this by training on vast amounts of unlabeled text data, such as
entire books, websites or comprehensive web crawls.
• By learning from this extensive corpus, LLMs develop the ability to predict the
next word or fill in missing words in a given context.

Examples of LLM’s
GPT-4 (Generative Pre-Trained Transformer 4): An autoregressive LLM developed by
OpenAI that has 2.5 trillion parameters and can perform various natural language
processing tasks with minimal fine-tuning.

BERT (Bidirectional Encoder Representations from Transformers) :

A bidirectional masked LLM developed by Google that has 340 million parameters and
can achieve state-of-the-art results on many natural language understanding tasks.

Google Gemini: A cutting-edge LLM created by Google DeepMind, featuring a

multimodal design capable of understanding and generating human language across
various formats. It is designed for superior performance on complex tasks, including
natural language processing, with state-of-the-art results in benchmarks and real-
world applications.

4
 3/19/2024

Three Types of LLMs

General-purpose LLMs Domain-Specific LLMs Proprietary LLMs

• These models are • These models are • These are LLMs you
trained on large trained on text from a can set up in your own
amounts of general specific domain like servers and train on
web text and aim to science, medicine or your company’s own
be useful for a wide law. They tend to proprietary data.
range of tasks. perform better for Eg: GPT-J / GPT-Neo
Eg: Chat GPT / Google domain-specific tasks.
Gemini / CLAUDE 3 / Eg: Bio-BERT / Sci-
Mistral Le Chat BERT

Multi-Modal LLMs

10

5
 3/19/2024

Generative AI for Modern

Marketing

Essential Generative AI
Tools Kit for Marketers
Module 2
1.Chat GPT
Getting to Know the 2.DALLE-E-3

Tools 3. Google Gemini

4.CLAUDE 3
5. Mistral AI (Le Chat)
6. Jasper AI
7.Grammarly PRO

11

Chat GPT
Initial Development and GPT-1 (2018): OpenAI introduced the first iteration of the Generative Pre-
trained Transformer, known as GPT, marking a significant step in natural language processing.

GPT-2's Controversial Release (2019): GPT-2, with 1.5 billion parameters, was initially withheld from
public release due to concerns over potential misuse for generating misleading information. It
demonstrated a remarkable ability to generate coherent and contextually relevant text passages, pushing
the boundaries of text-based AI.

GPT-3 and Widespread Adoption (2020): The release of GPT-3, featuring an unprecedented 175 billion
parameters, made headlines for its ability to perform a variety of tasks without task-specific training. Its
API provided developers access to its capabilities, leading to widespread adoption in industries ranging
from creative writing to customer service.

ChatGPT and Specialization (2022): ChatGPT built on the GPT-3 model to provide a more conversational
and user-friendly experience, becoming an integral tool for chatbots, virtual assistants, and educational
applications. Its fine-tuning for dialogue made it exceptionally capable of maintaining context in
conversations, a significant leap in user interactions with AI.

12

6
 3/19/2024

Practical Demo

• GPT 3 Vs GPT 4 – The Differences

• Enterprise Version Features
• Chat GPT Mobile App (Android/ iOS)
• Explore GPTs
• 3rd Party Plugins

13

Create Your Own GPT through My GPT

Chat GPT 4 comes with a customisable feature

called “My GPT”s.
Through My GPTs, you can create a personalised
chatbot specialized in a specific field, or you can
create a chatbot to remember a certain persona
and always respond in a consistent style.

14

7
 3/19/2024

Updating Custom Instructions with Chat GPT

Using custom instructions effectively with ChatGPT can significantly

enhance the outcomes of your interactions.
• Preferred Response Style: Indicate if you have a preferred style of
response.
• Background Context: Providing relevant background information, like
your field of expertise or specific interests, can help tailor the
responses more effectively.
• Examples of Ideal Responses: If possible, giving examples of responses
that you found particularly useful in the past can guide the AI to better
understand your expectations.

15

DALL-E (Generative AI for Image Generation by Open

AI)
Concept and Origin (Early 2020): DALL-E was conceptualized by OpenAI as a project to explore the
capabilities of AI in understanding and generating visual content. The project was named after the
famous artist Salvador Dalí and the animated character WALL-E, reflecting its focus on artistic and
creative image generation.

DALL-E Introduction (January 2021): OpenAI introduced DALL-E, a variant of the GPT-3 model
designed to generate images from textual descriptions.

DALL-E 2 Release (2022): OpenAI released DALL-E 2, a more advanced version with enhanced
resolution and realism in the images it generated.

DALL-E 3 : Released in 2023 with enhanced capabilities, and integrated into Chat GPT 4

16

8
 3/19/2024

Practical Demo

• Accessing DALL-E2
• Working with DALL-E3 on Chat GPT 4

17

CLAUDE 3 – A Chat GPT Alternative from Anthropic Inc.

Founding of Anthropic: Claude AI was developed by Anthropic, an AI safety

startup founded in 2021 by a group of former OpenAI employees.

Development and Training Approach: Anthropic developed Claude using a

technique known as Constitutional AI, which is designed to ensure that the AI
model remains helpful, harmless, and honest. This approach involves scaling
oversight and embedding ethical principles into the AI's architecture, setting
Claude apart from other AI models in terms of safety and reliability.

Features and Capabilities: Claude AI, especially the Claude 3 model, is noted for
its natural conversation abilities and has been designed to be "more steerable"
and less likely to produce harmful outputs compared to other models like
ChatGPT. It has undergone extensive training and incorporates feedback loops
to align its conduct with human values.

18

9
 3/19/2024

Practical Demo

• Creating a CLAUDE 2 Account

• Comparison with Chat GPT

19

Google Gemini

•Google Bard, created in 2022-2023, is a conversational AI model that leverages

the vast information available on the internet to provide informative, accurate,
and up-to-date responses to users.
•Google announced Bard, a conversational AI service in early 2023. Bard uses
Google's LaMDA language model to interact in a natural and intuitive way.
•Google Bard underwent extensive testing and refinement in 2023 to ensure its
conversational accuracy and ability to handle various queries.
•In the latest release, Bard uses Google's latest Gemini PRO AI language model
developed as a part of Google Deep Mind initiative, and now rebranded as
Gemini.

20

10
 3/19/2024

Practical Demo

• Creating a Gemini Account

• Using as a consultant for troubleshooting Google Marketing Tools

21

Practical Demo

• Mistral AI

22

11
 3/19/2024

Practical Demo

• Jasper AI

23

Practical Demo

• Grammarly PRO

24

12
 3/19/2024

Other Tools to Experiment With

Copy.ai
Microsoft Copilot
Midjourney
Stable Diffusion
Leonardo
getmagical.com

25

Homework 2

• Chat GPT Vs CLAUD Vs GEMINI

• Create your own user accounts on all three platforms. Play around with the
same prompts on three platforms and experience how differently or similarly
the three platforms respond. What are your personal views on the pros and cons
of each platform? Write a 300 to 500-word opinion.
• Use a Gen AI tool to fine-tune your writing but retain the original ideas.
• Share your response through the link available on your LMS. (Make sure you
submit through the ling for homework 2)
• This exercise carries 10% of your final evaluation.

26

13
 3/19/2024

What’s lined up next?

An Introduction to Prompt Engineering

* Having the right skills in writing effective prompts for Large Language Models will
set you apart from the rest of the people who are using these tools.

27

Professional Diploma in Digital Marketing

Amitha Amarasinghe
Co-Founder / CEO of APIDM

www.amisampath.com

[email protected]

Linkedin.com/in/amisampath Thank you!

28

14