An AI GRAPHIC Intro COURSE, for people with ZERO Knowledge.
For those friends, colleagues, and people asking me, ‘Where should I start if I want to learn AI with ZERO knowledge? A COURSE full of graphics, with very simple language, and each concept is followed by easy real-life examples. NEW! I made an update (3 FEBRUARY 2023) with 6 links to 5 of Google´s free AI essentials courses and 1 bonus video by Andrew Ng, see Index.
“This is the opportunity of a lifetime.
AI will bring more transformation than fire or electricity brought to our lives.”-
CEO of Google Spain, Google AI Zurich Executive Summit October 2023.
If you want to walk along with me on this long-life learning journey and enjoy learning together, subscribe to my publications.
I have meticulously developed a course that mirrors what I would have desired when embarking on my journey to learn about Artificial Intelligence (AI) from scratch. The material is conveyed in clear and accessible language, complemented by visuals tailored for individuals with a visual memory, a trait I share. Each concept is thoroughly explained and accompanied by multiple examples to facilitate a comprehensive understanding.
Moreover, in Part II of this course (refer to the INDEX), you’ll discover a concise summary of all the graphics employed throughout the entire course, condensed into just four slides.
Unlock the power of AI with one of the easiest AI Introduction Courses out there!
Dive into this step-by-step AI introductory COURSE, meticulously crafted to demystify essential jargon through captivating visuals — perfect for those with a photographic memory.
Immerse yourself in a condensed version of Google’s AI course ( video and readings), Chat GPT documentation, and AI engineer's sources, infused with graphics, and simple language, and each concept is explained with multiple and easy examples.
Don’t miss out on this course, is your shortcut to AI essential rounded knowledge!
🔗 Find ALL links to sources at the end of this Course.
📌 INDEX
🪣 INTRO.
. Who is this essential AI course for?
. What sets this course apart?
🪣 PART I. FULL AI INTRODUCTION COURSE.
EPISODE 1. What is AI?
EPISODE 2. What is Machine Learning?
EPISODE 3. What is Deep Learning?
EPISODE 4. What is Generative AI (Gen AI)?
EPISODE 5. What are Large Language Models?
🪣 PART II. ONLY 4 SLIDES. AI FULL GRAPHIC COURSE
SLIDE 1. What is AI, What is Machine Learning?
SLIDE 2. What is Deep Learning?
SLIDE 3. What is Generative AI?
SLIDE 4. What are Large Language Models?
🪣 END NOTE.
🪣 ANNEXE.
ANNEXE 1. About the author.
ANNEXE 2. SOURCES.
🪣 🤩 GOOGLE CLOUD, 5 FREE TRAININGS ON GEN AI (UPDATED 3 FEBRUARY 2023), and 1 bonus video by Andrew Ng.
INTRO.
Who is this essential AI course for?
This AI INTRODUCTION VISUAL COURSE, is tailored for anyone who aims to comprehend technical AI concepts without a Tech background, I made this AI COURSE for those friends, colleagues, and people who are asking me where I should start if I want to learn AI with ZERO knowledge, hope you ALL like it, I made this with LOVE.
I have served as a professor at ISDI, one of the most renowned digital business schools in Europe, for more than 4 years, where I taught Digital Strategy, Data Analytics, and CRM Strategies to C-level leaders. Leveraging this teaching experience, I have meticulously crafted this step-by-step guide, intending it to serve as an introductory class for anyone interested in AI.
If you’re a business leader, or a person with ZERO knowledge in AI, aspiring to grasp the fundamentals of artificial intelligence in simple language and through visuals, then this GUIDE is for you.
What sets this AI essential course apart?
🚀 What sets this course apart is its unique approach to simplifying AI technicalities by,
1. Using graphics, to those who learn effectively through visual associations, as is my case, and have a photographic memory.
2. Simple language.
3. Every concept followed by real examples.
4. Not only Google AI Course but from other VERY important sources, like Chat GPT, AI engineers…all the sources are linked at the end of the course.
5. All the graphics of this Full AI introduction course are summarised, in only 4 slides.
Initially, by the time I was digging into this AI Google Course, I had reservations about potential abstraction in content. However, my concerns were alleviated as the foundational concepts not only exceeded my expectations but also enhanced my practical knowledge with tools like Chat GPT and Google Bard.
The course not only helped me but also dispelled several misconceptions I held about AI, machine learning, and large language models.
This visual AI course strives to offer a comprehensive visual aid, encapsulating the essence of acquired concepts for efficient and accessible understanding.
It is time to start, and let’s go with a fundamental question:
🤖 What is artificial intelligence?
To my surprise, I wasn’t aware that AI is an entire field of study.
If you find yourself in a similar position, fear not — we’ll navigate through these concepts together.
EPISODE 1. What is AI?
💡Artificial Intelligence is a whole field of Study. Here’s a list naming various fields of science:
- Physics
- Chemistry
- Biology
- Astronomy
- Geology
- Meteorology
- Mathematics
- Artificial Intelligence (AI) … and many more.
Inside AI, there’s something called Machine Learning, which is like a smaller part.
🗒️ Example- Artificial Intelligence.
Let’s imagine Artificial Intelligence as a big library full of books. Inside this library, there’s a section dedicated to a smaller topic known as Machine Learning. It’s like saying the whole library is AI, and the Machine Learning section is a specific area where we focus on a particular kind of learning for computers.
Then, within Machine Learning, there’s Deep Learning, which is even more specific.
Deep Learning includes things called Generative Models, and among them are Large Language Models.
It’s like saying AI is the big umbrella, and Machine Learning, Deep Learning, Generative Models, and Large Language Models are all different layers underneath it.
EPISODE 2. What is Machine Learning?
💡 Machine learning uses input data to train a model that can make predictions based on unseen data. Machine Learning is a Sub-type of AI.
In Machine Learning, we have Supervised learning models that use labeled data, while Unsupervised learning models use unlabeled data.
Supervised models can predict outcomes based on historical data, by using labeled data, while Unsupervised models identify patterns in raw data models that work with unlabeled data.
🗒️ Example Machine Learning.
Imagine teaching a computer like you teach a friend. In “supervised learning,” you give the computer labeled examples, like saying, “This is an apple, that’s a banana.” It learns by comparing these labeled examples.
Now, in Unsupervised learning, you don’t label everything. You let the computer find patterns on its own. It’s like saying, “Sort these fruits without telling you which is which.” The computer figures out groups based on similarities. Unsupervised problems are all about discovery, about looking at the raw data and seeing if it naturally falls into groups.
🗒️ Example Let’s go into a practical real business example: As a real example If you train a model using ZARA’s coat sales data, you can then use that model to predict how well a Mango coat would sell based on ZARA’s sales data.
In a nutshell, Machine learning is a program that uses input data to train a model. This trained model can then make predictions based on data it has never seen before.
EPISODE 3. What is Deep Learning?
Inspired by the human brain, they look very similar:
💡Deep Learning is a brain, Deep learning is a type of machine learning that utilizes artificial neural networks, that have many layers of neurons, which allows them to learn more complex patterns than traditional machine learning models.
Deep learning, in a Semi-Supervised Learning model, is trained on a small amount of labeled data and a large amount of unlabeled data.
🗒️ Example. Let’s go into a practical real business example:
Banks and financial services use deep learning models to detect fraud.
The bank spends some time tagging or labeling 5% of transactions as either fraudulent or not fraudulent, leaving the remaining 95% of transactions unlabeled. This is because they lack the time or resources to label every transaction. The magic happens when the deep learning model uses the 5% labeled data to grasp the basic concepts of the task — identifying what transactions are good or bad.
The model then applies these learnings to the remaining 95% of unlabeled data, creating a new aggregate dataset. With this, the model makes predictions for future transactions.
This is awesome, but we’re not done yet, because Deep learning can be divided into 2 types:
- Discriminative models.
- Generative models.
Discriminative models learn from the relationship between labels of data points and only can classify those data points as fraud, not fraud ( like the example we talk about above).
🗒️ Example.
You have a bunch of pictures or data points you purposefully label some of them as cats, and some of them as dogs, a discriminative model will learn from the label cat or dog and if you submit a picture of a dog it will predict the label for that new data point a dog.
We finally get to Generative AI unlike discriminative models, where generative models learn about the patterns in the training data after they receive some input.
So now we are ready to go to fully understand GEN AI.
EPISODE 4. What is Generative AI (Gen AI)?
Finally, we get to generative AI.
💡Gen AI, Unlike discriminative models, generative models learn about the patterns in the training data after they receive some input. Is a type of artificial intelligence technology that can produce various new types of content, including text, imagery, audio, and synthetic data.
Gen AI marks a significant leap in the evolution of artificial intelligence. Unlike its predecessors, which were specialized in specific tasks, GEN AI aims to mimic the broad learning capabilities of the human mind.
Imagine an AI system that doesn’t just excel in one area but can adapt and learn across various domains. GEN AI is designed to be versatile, enabling it to tackle diverse challenges and problem-solving scenarios. This generalization empowers it to learn from different experiences and apply that knowledge in novel situations.
One of the defining features of GEN AI is its ability to connect the dots between seemingly unrelated tasks, much like how humans learn from various experiences to make informed decisions. It’s not just about memorizing data; it’s about understanding the underlying principles and applying them creatively.
🗒️ Example. Gen AI
When given a text prompt, they generate something new based on the patterns they just learned. Going back to the animal example, the pictures or data points are not labeled as ‘cat’ or ‘dog.’ So, a generative model will look for patterns. These data points all have two ears, four legs, and a tail, like dog food, and bark. When asked to generate something, the generative model creates a completely new image based on the patterns it just learned, based on the identified characteristics.
There’s a super simple way to determine if something is generative AI or not
It is not Gen AI, if>
if the output is a number a classification spam, not spam or a probability.
🗒️ Examples are not GEN AI
These examples represent outputs that don’t involve generating new content or creative outputs, making them “Not Gen AI.”
1. Number Output: 42 (Just a numerical value)
2. Classification Output: “Cat” or “Dog” (Classifying something without generating new content)
3. Spam Classification: “Spam” or “Not Spam” (Determining if an email is spam or not)
4. Probability Output: 0.75 (Assigning a probability value to an event)
It is Gen AI, if>
when the output is natural language, text, speech image, or audio generative AI generates new samples that are similar to the data it was trained.
🗒️ Examples are GEN AI
In these examples, Generative AI is involved in producing content that exhibits creativity and originality.
Here are examples of when we can determine that something has been generated with Generative AI:
1. Natural Language Text: A paragraph of text that reads like it was written by a human.
2. Artistic Creation: A unique piece of artwork or image generated based on learned patterns.
3. Speech Synthesis: An audio clip that sounds like a human speaking, created by the model.
4. Novel Content: A completely new and original piece of writing or composition.
5. Image Synthesis: Creating realistic images of non-existent objects or scenes.
In a nutshell, generative AI generates new samples that are similar to the data it was trained on.
Let’s see the 5 main different TYPES of GEN AI we have and also some tools, further than Google solutions.
- TEXT to TEXT.
In text-to-text models, the input and output are both treated as text, allowing the model to handle diverse tasks such as translation, summarization, question answering, and more.
🗒️ Examples tools text to tex.
GPT-4 and Google Bard.
Bard and Chat GPT, are powered by a large language model, which is a type of machine-learning model, pre-trained on a massive amount of diverse text data.
2. TEXT to IMAGE
The different generative AI model types like Chach GPT, Google Bard, or BING, include text-to-image models. Bing GEN AI model, generates images using DALL-E.
🗒️ Examples tools text to Image.
Mid-Journey, DALL-E2, and Stable Diffusion can not only generate images but some of them also edit.
3. TEXT to VIDEO.
Last Week, Yariv Google’s Senior Director Conversational Generative AI Lead, whom I met in Zurich ( I was invited by Google’s AI Zurich Executive Summit), introduced VideoPoet.
A recent wave of video generation models has burst onto the scene, in many cases showcasing stunning picturesque quality. One of the current bottlenecks in video generation is in the ability to produce coherent large motions. In many cases, even the current leading models either generate small motions or, when producing larger motions, exhibit noticeable artifacts.
To explore the application of language models in video generation, we introduce VideoPoet, a large language model (LLM) that is capable of a wide variety of video generation tasks, including text-to-video, image-to-video, video stylization, video inpainting and outpainting, and video-to-audio. +More info
🗒️ Examples of tools text to video:
⏯️ Find in this video, how VideoPoet generates videos with audio.
Also, Meta has its text-to-video AI tool, and you can check out the AI text-to-video Tool Runway.
⏯️ In this video, the author covers the Runway’s video tool.
4. TEXT to 3D
The AI text-to-3D model allows you to generate 3D models from text descriptions. This is useful in fields like architecture, design, or gaming.
🗒️ Examples of tools text to 3D:
- Ponzu: Best for developers and designers.
- Meshcapade: Best for e-commerce-ready models.
Find more examples: 10 Best AI 3D Generators 2023
4. TEXT to TASK.
Text-to-task AI tools refer to applications or systems that can interpret and convert natural language text input into specific tasks or actions. These tools utilize natural language processing (NLP) and machine learning techniques to understand and extract actionable information from textual data.
NLP is a field of artificial intelligence that focuses on the interaction between computers and humans through natural language. Text-to-task AI tools leverage NLP techniques to comprehend the meaning and context of the input text.
🗒️ Example tool text to task:
I am registered in the Google Cloud AI Learning Center, and I have tons of courses to address. However, I am not clear on what kind of AI courses are more suitable for me. Therefore, I reached out to Bard:
“Hello Bard, I am registered in the Google Cloud AI Learning Center, I am not a developer, I am an e-commerce director, which AI courses should I take? ( see image)”
and this is its answer very well structured about the courses I Should take:
this is its full answer:
Sure, here is a list of AI courses from Google Cloud that might be relevant to you as a digital e-commerce director:
Introduction to AI — This course provides a high-level overview of artificial intelligence, including its principles, applications, and ethical considerations.
AI for Decision-Making — This course teaches you how to use AI to make better decisions in your business. You will learn how to build predictive models, use natural language processing to analyze customer feedback, and optimize your marketing campaigns.
AI for Product Innovation — This course shows you how to use AI to develop new and innovative products and services. You will learn how to use machine learning to identify trends in customer behavior, use natural language processing to understand customer needs, and design products that meet those needs.
AI for Marketing Automation — This course teaches you how to use AI to automate your marketing tasks and improve your marketing ROI. You will learn how to use machine learning to segment your customer base, use natural language processing to personalize your marketing messages, and use chatbots to provide customer support.
AI for Customer Service — This course shows you how to use AI to improve your customer service. You will learn how to use chatbots to provide 24/7 support, use natural language processing to identify customer issues, and use machine learning to predict customer churn.
These courses are designed for people with no prior experience in AI. They are taught by Google Cloud experts and cover a wide range of topics, including the basics of AI, machine learning, natural language processing, and deep learning.
I hope this helps!
and if I wonder how much time it will take me to study all of them this is the answer:
Also gives me and Google sheet with the calculations.
and also you can ask the TASK to export them to your Google Calendar!
EPISODE 5. What are Large Language Models?
💡 Moving over to large language models, let’s remember that LLMs are also a subset of deep learning, and although there is some overlap LLMs and GEN AI, are not the same thing.
An important distinction is that Large Language Models are generally pre-trained with a very large set of data and then fine-tuned for specific purposes, and are a class of Foundation Models, or FMs, in artificial intelligence.
These Foundation Models are large-scale AI frameworks trained on a broad spectrum of data and tasks, capable of adapting to a diverse array of applications with task-specific fine-tuning.
What does that mean?
🗒️ Example LLMs
Imagine you own a pet dog that has been trained with basic commands such as sit, come, down, and stay.
This dog is well-behaved and versatile in general. However, if that same well-behaved dog undergoes training to become a police dog, guide dog, or hunting dog, it needs specialized training to excel in that specific role. A similar concept applies to large language models (LLMs).
Initially, they undergo pre-training to tackle common language challenges like text classification, question answering, document summarization, and text generation.
Subsequently, these LLMs are fine-tuned using smaller datasets specific to industries such as retail, finance, healthcare, entertainment, and other fields, allowing them to address distinct problems in the real world.
🗒️ Examples Real life LLMs Business applications
Example (1) A Big Retailer may leverage a pre-trained LLM from a major tech company and fine-tune the model with its proprietary sales, margin, and marketing data, enhancing sales accuracy in interpreting X-rays and other medical tests.
Example (2) consider a hospital employing a pre-trained large language model sourced from a major tech company like Chat GPT, Google, or Meta. The hospital then refines this model using its proprietary medical data, aiming to enhance diagnostic precision in interpreting results from X-rays and other medical tests.
This mutually beneficial scenario epitomizes a win-win situation, as large corporations invest substantial resources, in the order of billions, to craft versatile, general-purpose large language models.
🚀 A WIN- WIN.
This symbiotic relationship is advantageous for both parties; enabling them to leverage sophisticated AI capabilities without the need for extensive developmental investments. Large companies can invest billions in developing general-purpose LLMs, which they then provide to smaller institutions like retail companies, banks, and hospitals. These smaller entities may lack the resources for developing their own LLMs but possess valuable domain-specific datasets that empower them to fine-tune these models for their specific needs.
🗒️ Example TOP LIST of companies and their corresponding LLMs:
1️⃣ OpenAI’s ChatGPT
ChatGPT is a GPT variant tailored for conversational responses. It operates through understanding and generating human language and can be adapted for numerous conversational tasks.
The language models mentioned are examples of the latest generation of Foundation Models in AI and natural language processing (NLP).
2️⃣ Google’s Bard
In response to ChatGPT’s success, Google’s Bard is built on advanced language modeling technology, utilizing variants of models like BERT or LaMDA ( Google’s conversation technology).
3️⃣ Meta’s LLaMA 2
The LLaMA (Language Model Meta AI) suite comprises variously sized language models by Meta AI.
4️⃣ AI.com’s Grok
Launched by Elon Musk, Grok is an FMs detailed in my coverage last Saturday.
5️⃣ Anthropic’s Claude
Claude is a language model developed as a FM, focusing on comprehensibility, usefulness, and safety in AI-human interaction.
6️⃣ AI21 Labs’ Jurassic-2
Jurassic-2 is a potent language model that caters to various natural language processing tasks, embodying the trend of large models adaptable for specific applications.
7️⃣ Cohere’s Command
Cohere offers natural language processing APIs powered by large-scale language models.
Each of these models shares similar architectures and has been trained on vast amounts of text data.
PART II. ONLY 4 SLIDES. AI FULL GRAPHIC COURSE
SLIDE 1. What is AI, What is Machine Learning?
SLIDE 2. What is Deep Learning?
SLIDE 3. What is Generative AI?
SLIDE 4. What are Large Language Models?
✨ END NOTE
I’ve meticulously crafted a course that reflects the kind of learning experience I would have cherished when beginning my AI journey from scratch.
This course is thoughtfully designed with a tender touch, taking into consideration the needs of its audience.
It unfolds in straightforward language, complemented by visually appealing graphics tailored for those, like myself, who thrive on visual memory.
Each concept is not just explained but brought to life with a wealth of examples, ensuring a thorough and enjoyable learning experience.
I want to contribute to humanity by embracing AI because I truly believe in the huge impact of AI in our lives.
I’ve worked on AI my whole life because I’ve always felt would be the most beneficial and consequential Technology For Humanity.-
Demis Hassabis, CEO of Google DeepMind.
If you want to walk along with me on this long life learning journey and enjoy learning together, subscribe to my publications.
THANKS for Learning together, IKUZO.
ANNEXE 1. ABOUT THE AUTHOR.
🛒 LINKEDIN PROFESSIONAL PROFILE.
. 🤖 Mar Fernández as a AI Speaker:
. 🎙️ Mar Fernández recent AI interviews:
. ✍🏻 Others Mar’s AI Short Publications:
ANNEXE 2. SOURCES.
🙌🏻 Google FULL course:
https://www.youtube.com/watch?v=N3sMj-sBcLg&t=256s
🙌🏻 Other great sources:
https://www.youtube.com/watch?v=-a6E-r8W2Bs
https://www.linkedin.com/in/marfernandezparra/recent-activity/all/
https://www.eweek.com/artificial-intelligence/best-ai-3d-generators/
https://www.youtube.com/watch?v=cw_lQ57qWwg
https://blog.research.google/2023/12/videopoet-large-language-model-for-zero.html
https://www.youtube.com/watch?v=0EGogz-tCls
https://www.youtube.com/watch?v=Yq0QkCxoTHM&t=10s
🪣 GOOGLE CLOUD, 5 FREE TRAININGS ON GEN AI (UPDATED 3 FEBRUARY 2023)
1️⃣ Intro to Generative AI
https://www.cloudskillsboost.google/paths/118/course_templates/536
2️⃣ Intro to Large Language Models
https://www.cloudskillsboost.google/paths/118/course_templates/539
3️⃣ Intro to Responsible AI
https://www.cloudskillsboost.google/paths/118/course_templates/554
4️⃣ Generative AI Fundamentals
https://www.cloudskillsboost.google/paths/118/course_templates/556
5️⃣ Responsible AI — Applying AI Principles
https://www.cloudskillsboost.google/paths/118/course_templates/388
Bonus video:
6️⃣ Andrew Ng: Opportunities in AI — 2023
Dr. Andrew is the Managing General Partner of AI Fund, founder of Deep Learning AI and Learn, and Chairman and co-founder of Coursera. He is an Adjunct Professor of Computer Science here at Stanford. Previously, he had started and led the Google Brain team, which played a crucial role in helping Google adopt modern AI. He was also the Director of the Stanford AI Lab. About 8 million people, or one in one thousand persons on the planet, have taken an AI class from him. Through both his education and his AI work, he has changed numerous lives
