AI for Beginners: Breaking it Down in Simple Terms
Introduction
Artificial Intelligence. Everyone's talking about it, and it's disrupting every aspect of our lives like never before. It's changing how we learn, how we work, and how we interact in our everyday lives. And it's here to stay.
So, if this groundbreaking tech has piqued your interest but feels a bit intimidating, don't worry! This post is your starting point for understanding Artificial Intelligence, and we're kicking it off with the basics.
What is AI?
Okay, let's keep it simple.
At its core, AI is the science of training machines to perform tasks that typically require human intelligence. It's all about programming computers to think like we do - to learn, reason, perceive, infer, communicate and make decisions.
But instead of doing a manual job or simply executing programmed instructions, these machines actually learn from experience and adjust to new inputs, just like our human brain.
Think about things you do every day - checking your emails, Googling something, using a maps app to navigate, or writing social media posts. All these activities involve a measure of intelligence, right? Now, AI is positioned and designed to replicate this human intelligence.
Let's unpack this a bit more with this definition from Christopher S. Penn: "If a task is repetitive and involves creating something (like a weekly recap email to your boss), it's a good candidate for AI to assist or outright do."
Essentially, if a task involves something we humans would consider 'smart' but can also be somewhat repetitive, AI can likely help automate that task.
AI can be divided into two broad types:
Narrow AI, which is designed to perform a specific task, like voice recognition.
Example: Siri or Alexa. These AI systems can interpret and respond to voice commands, play your favourite songs, tell you the weather, or set reminders for you. However, they operate within a limited context and cannot perform tasks beyond their programming.
General AI, which can theoretically perform any intellectual task that a human can do.
Example: A system like the one in the movie 'Her'’ where the AI can carry on fully human-like relationships, interpret complex emotional states, and make decisions based on nuanced, contextual information.
Let’s go further…
We can take our understanding of AI even further by categorizing it into three main groups - prediction, classification, and generation.
Prediction involves making future decisions based on past data. It's like your weather app forecasting if it's going to rain tomorrow based on today's climate conditions.
Example: Streaming services like Netflix or Spotify predict what movie or song you'd want to watch or listen to next based on your past behaviour. Another example is your GPS predicting your ETA based on current traffic conditions.
Classification is about categorizing data into predefined labels.
Example: Think of your email spam filter that classifies emails as 'spam' or 'not spam'’
Generation is creating new data or pieces of information. This is where AI gets really creative, and it's what we'll dive deeper into next.
Example: Tools like OpenAI's ChatGPT, which can generate human-like text, or Midjourney, which creates original pieces of art in various styles, are great examples of this.
In essence, AI is an evolving field that's striving to create intelligent machines that work and react like humans, but faster and without getting tired. It's a truly transformative technology that's set to revolutionize how we live and work. And understanding its basics is the first step in this exciting journey.
What is Generative AI?
Generative AI is the cool kid in the AI world right now, and It's all about creating. It can create entirely new things or take what's already there and make it better. These "things" can be written work, like emails or strategic plans, songs, images, and more.
There are two main categories of Generative AI, and which one you pick depends on what you're looking to create.
Transformers are mostly found in the world of language generation. They're part of software called large language models (the "T" in ChatGPT stands for Transformer). Transformers can write new texts like emails, answer questions, summarize huge chunks of text, rewrite text, extract info from texts, and organize text.
Diffusion models are the artists of the AI world, generating pictures and images. Services like Midjourney use this model to create brand-new images based on a written prompt. It can also modify these images based on your directions, tweaking style, colours, and more.
As you gear up to work with AI, you'll first need to decide if you're more of a word person or a picture person. Every new generative AI tool out there specializes in either transformers or diffusers.
Where Do I Start?
This is where most folks start to feel overwhelmed.
With thousands of new AI companies popping up all over the place, it can be hard to know where to start. So, let me make it easy for you.
Generating Text
OpenAI's ChatGPT is your best bet for diving into large language models. It's got a free version, it's one of the top tools out there right now, and it's super easy to get started.
Prompt Engineering
Once you're all set up with ChatGPT, the first skill you'll want to master is 'prompting'’ This basically means learning how to chat with the program to get it to do what you want it to do.
Here is the template I like to follow:
Act As A
Topic
Audience
Format
Tone
Goal
Context/Background
Ask me questions
Prompting In Action:
Let’s try this out by imagining we are a school principal getting ready to send a communication home and fill in the template:
Act As A: A respected and approachable school principal with strong communication skills. The principal is in charge of an elementary school and is known for fostering a close-knit school community.
Topic: Changes to school start times that have been carefully considered and decided upon for the next academic year.
Audience: The parents and caregivers of the students who attend the elementary school. These individuals value clear and transparent communication and are heavily invested in their children's education.
Format: A detailed and thoughtful letter to be sent home with students and shared via email. The letter should follow professional formatting for such communications, with a clear structure, headings for each section, and a friendly closing.
Tone: The letter should strike a balance between professional and friendly. While the message being delivered is of a formal nature, the language used should convey warmth, understanding, and care.
Goal: The primary goal is to inform parents about the upcoming changes in the school's start times, explaining why this decision was made. The letter should also address potential concerns proactively, reassuring parents about the benefits of this change and demonstrating that the school's administration has a comprehensive plan to ensure a smooth transition.
Context/Background: The decision to alter school start times has been made after a thorough review of various studies indicating benefits for children's academic performance and overall well-being. Furthermore, this decision has been influenced by feedback received from parents and teachers and logistical considerations. (Here, include the specific details about the new start times, why the decision was made, and any other relevant context.)
Ask me questions: Before generating the content, the AI should ask about specific concerns parents may have, any support measures the school has in place for the transition, and any other context that would be important for the parents to know.
So, the improved prompt would look something like this:
"Act as a respected and approachable school principal in charge of an elementary school known for fostering a close-knit school community. You are preparing a detailed and thoughtful letter addressing the parents and caregivers of our students about significant changes to school start times for the next academic year. The tone should be both professional and friendly. The goal of this communication is to clearly explain the reasons behind this change, proactively address potential concerns, and assure parents that plans are in place for a smooth transition. The decision to change the school start times was influenced by academic research, parent and teacher feedback, and logistical considerations. (Insert specifics about the start times and why the decision was made here). Please ask me questions to gather more information before we begin drafting the letter."Prompt: “Act as a school principal of an elementary school with strong communication skills. You are creating a communication for parents about changes in school start times in the format of a letter. The letter will be friendly and informative and use professional language. The goal of this letter is to proactively address their concerns and ensure a smooth transition to the change. Here is a little more background (insert specifics about the start ties and why the decision was made here). Please ask me questions before we get started.”
Heads up: don't input any sensitive or confidential information into ChatGPT, as it might be used to train the model.
As you get used to ChatGPT, you'll notice that the text it generates might seem a bit bland and average and not always spot-on. You'll need to keep working on your prompts, adding more details on tone, style, word choice, and so on, to get the results you want. But don't worry; we'll cover more on that in future posts.
Generating Images
Unleashing your creativity with image generation is a whole new level of excitement. Personally, I'm a fan of MidJourney. The idea is similar to ChatGPT; by entering what you envision a picture to be, the system generates an image based on your prompt, providing you with the opportunity to refine it.
Here’s an example of an image prompt:
“A bustling classroom of the future: diverse group of students engaged in interactive, virtual reality lessons. Each student wears a high-tech headset and uses tactile holographic displays to explore vibrant, three-dimensional educational content. The walls are covered in digital art created by students, while a friendly robot assistant helps a small group in the corner. Style: A vibrant and detailed digital illustration with a hopeful, optimistic tone.”
And what we end up with:
And the magic doesn't stop there. You can refine and fine-tune the image according to your wishes. Make it darker, brighter, more detailed...the possibilities are endless, and the control is entirely in your hands.
PRO TIP: If you're finding it challenging to craft the perfect image prompt, don't sweat it. ChatGPT can be your personal assistant in this aspect too. Just ask it to help you come up with an image prompt that you can use in MidJourney. It's like having an art director and a painter rolled into one!
Conclusion
And there you have it! An ultra-high-level whirlwind tour over the landscape of AI. By now, you should have a basic understanding of what AI is, its different types, and a practical guide on how to use it, particularly focusing on language and image generation.
But remember, this is just the tip of the iceberg! The world of AI is vast and ever-evolving. Stay tuned for future posts where we'll delve deeper into the intricacies of AI and navigate this fascinating terrain together.
So, whether you're just dipping your toes into the waters of AI or planning to dive deep into its complexities, remember to enjoy the journey. Explore, experiment, and ask questions. We're all learners here in this brave new world of AI!
