Explore AI Programmatically

0:00Hello and welcome to the AI-Agentic Coding ChatGPT Cursor, Clod, and Co-Pilot course.

0:08I'm excited that you're here.

0:10When most people think of AI, they think of a chatbot window that spits out fluent, correct-sounding

0:16answers, helpful but not always correct, and it hides the bigger picture.

0:22When it comes to programming, the power of AI can be a bit overwhelming, and at the same

0:27time it can be welcoming to a wide audience, anyone from software engineers to complete

0:33beginners.

0:34Before we talk about the scope of this course, I did want to say that this is a beginner

0:39course.

0:40Meaning, if you have some experience programming and want to start using AI to increase your

0:45existing programming skills, then this is the course for you.

0:49But it's not entry-level.

0:51And by that I mean, if you don't have a programming background, this might not be the right course

0:56for you.

0:57But no worries, I have a course just for you on VibeCoding, and I'll explain that in just

1:02one second.

1:03First, let's check out what's included in the course, what's in scope and what's not.

1:09This course is about understanding the basic prompt engineering skills for programming.

1:14For example, when to use one model over the other for programming.

1:19And we'll talk about how to install and use different AI coding tools.

1:23So keep in mind that this is a mini-course.

1:26I'll show you that this is part of a series of AI mini-courses.

1:30So we're not going to get really deep into any one topic.

1:34But if you want me to create a course for, let's say, cursor or cloud code, just leave

1:39it in the comments.

1:40We read those comments and those will give us an idea of what courses that you're more

1:44interested in.

1:45This is more of a mini-course where we explore all of the different software, how they work,

1:51This is a mini crash course on prompt engineering as it applies to programming.

1:56We're not going to learn about programming in general.

1:59I have a course on VibeCoding instead.

2:02I'll show you that in the next slide.

2:04And there are no tutorials on how to build AI chatbots.

2:07So I did use the word agentic, which is agents, AI agents, but we're not building those.

2:13We're going to use existing programming AI agent tools.

2:17Again, we're not going to use APIs in this course because this is a mini-course.

2:22But if you want to go deeper into any one of these and include APIs, please let us know

2:27in the comments.

2:28So when I refer to APIs, I mean, we're not going to use computer to computer connections

2:32to perform any kind of AI programming other than that built into tools like Cursor or

2:40GitHub Copilot and so on.

2:42And like I said, this course is part of an AI mini-series.

2:46And this is the course that we're on now.

2:48However, if you don't have any programming experience, I definitely recommend the AI

2:54VibeCoding and Security course.

2:56Here's why.

2:57First of all, if you don't have any programming experience, you can ask ChatGPT to give you

3:02code and you can paste it somewhere or execute it and create a game.

3:08And some people are actually creating businesses doing this process because, well, you have

3:13business ideas, you maybe have great ideas, but you don't know how to program.

3:18And VibeCoding allows you to ask the agent or the AI tool to program for you.

3:24When you do that, there are a bunch of security issues that come up and that's what we're

3:28going to address here.

3:29It's like a crash course on how to do VibeCoding safely.

3:32And here, this is for software engineers or people that have a programming background.

3:37Let's say you finish this course.

3:39This is a good next course to take.

3:41But if you're a software engineer or you have a programming background, this is a good

3:45course to learn how to use the different tools like ChatGPT, Cursor, Clod, Copilot.

3:52And one quick caveat, ChatGPT, we don't really use that for agentic coding, but it does have

3:58a purpose and we're going to use it to learn different concepts.

4:02But ultimately, it's not for agentic programming.

4:05It's more for planning.

4:06More on that in the course.

4:08And so what are all these other courses?

4:11So here, prompt engineering is directly tied to this course.

4:14I'm going to give you a quick crash course in prompt engineering as it applies to programming.

4:20But I highly recommend taking this course because we all talk to the chatbot in different

4:25capacities.

4:26It's not always for programming.

4:28And when you do that, it will hallucinate, make things up and drift.

4:32If you don't know how to talk to it, then it makes it a lot more difficult.

4:36And in this first skill, we're going to talk about AI programmatically and explain the

4:41different types of software and give you a working mental model of how AI works, which

4:46lays the foundation for prompt engineering.

4:49But I will give you a crash course here in case you already have a little bit of a background.

4:54And if you already know prompt engineering, you can go straight to the AI coding tools.

4:58If you're working with Microsoft and you're curious about Microsoft Copilot, which is

5:02different than GitHub Copilot, this is a great course to take.

5:05If you're a business leader or an executive or you want to launch your own AI product,

5:10this is a good one to take as well.

5:11Now here we have two courses that are very similar, but for different audiences.

5:16AI compliance, privacy and copyright is a very important subject.

5:21And this is the technical version here for developers.

5:24And this is for a non-technical audience, HR, legal and IT.

5:28And finally, we have AI for productivity.

5:31This is for anybody.

5:32And it's very similar to what you might do here, but it's non-tool specific.

5:37So it's tool agnostic.

5:39This one up here, this is all Microsoft.

5:42And a little bit about me.

5:43My name is Jonathan Barrios.

5:45I teach data science, machine learning and deep learning.

5:48And all of this can fit under the umbrella of data science.

5:52But now that we're in 2025, it's really now called AI data science.

5:58I blend data science and AI engineering into one field.

6:02And I've also worked with AI and built software as a service applications, many different

6:07AI applications and even streaming platforms.

6:11Last but not least, I've been teaching for a while at top platforms from Treehouse to

6:15Thinkful and Chegg.

6:17And now I'm happy to be part of the CBT Nuggets team.

6:20And if you want to find me online, you can definitely do so at LinkedIn here at Jonathan-Barrios-AI.

6:27And my website where I occasionally publish different posts.

6:31There's one there on vibe coding, if you want to read that.

6:34And this is an important one.

6:36People have mixed feelings about X.

6:38The reason that I'm there is because that's where the AI community is.

6:42If you don't want to sign up for a newsletter to keep up with all of the developments in

6:46AI, XAI is a really easy way to do that.

6:49Furthermore, you can directly communicate with the greatest minds in AI directly on

6:54this platform.

6:55I do not like social media personally.

6:57I don't use it.

6:59However, in this case, it's kind of working well.

7:03So the way that I do this is I follow only people that talk about AI and then I ignore

7:07everybody else.

7:08That's the way that I do it.

7:10And if you want to follow me, I promise it's about AI data science.

7:14And that's my handle, AI underscore data underscore science at X.com.

7:19All right.

7:20Now, moving on to our first skill here.

7:23We're going to zoom out and look at AI as a whole.

7:27And we're going to think of it in terms of software, meaning what is AI programmatically.

7:32We'll discuss different types of software.

7:34First, we'll explore what is AI because it's not one thing.

7:40Many people think that it's the chat bot.

7:42That's actually generative AI and that's just one subdomain in AI.

7:47So if that's what you think, then this is going to be very illuminating.

7:50But if you have a good handle on this, then it might be interesting to learn about the

7:54different types of software.

7:56For example, software 1.0, 2.0 and 3.0.

8:011.0 is traditional.

8:03This is machine learning and deep learning, and this is generative AI.

8:07And there's a good reason why I explain it this way.

8:10In short, we're framing it this way to keep you from overcomplicating things.

8:15There are many tasks that still belong to traditional programming.

8:19These are simple rules, but here we're going to build a deep neural network in this skill.

8:24And these are models that you build.

8:25This is machine learning and deep learning.

8:28And here, this is when we talk to a chat bot.

8:31That's generative AI.

8:32And that's what we use for agentic programming, right?

8:36We're using an AI agent, a chat bot, to do programming together using some kind of a tool.

8:42And then we'll talk about guardrails.

8:44Now prompt engineering, just the idea about human in the loop.

8:49AI can't do it by itself.

8:50It needs us right now.

8:52And this is a very important concept.

8:54And we can use this as a safety guardrail.

8:57And then finally, we have a challenge.

8:59These challenge sections are for you to test your knowledge, meaning identify any knowledge

9:05gaps before moving forward.

9:07That way, every successive skill will make sense and you'll get the most out of it.

9:12Let's say that you already know what AI is.

9:14You know what software 1.0, 2.0 and 3.0 is.

9:18And furthermore, you know prompt engineering, well, you can just skip those sections, right?

9:23Definitely take the challenge to see if you understand that skill.

9:26And if you do, move on to the next one.

9:28And that's totally fine.

9:30Here's why.

9:31I made this very accessible.

9:33So you could be a beginner or you could be intermediate or advanced.

9:36It really depends on where you're coming from.

9:39So if you just need to learn those tools, then the last three skills are going to be

9:44just for you.

9:45If you need to ramp up to those first three skills, I've included what is AI programmatically,

9:51prompt engineering, and then some fundamentals and how to use chat GPT for development.

9:56Definitely don't use it for code.

9:58We'll talk about that in the third skill.

10:00So that should give you an idea of what we're going to do in this skill.

10:03And I'll see you in this very first video here.

10:06What is AI?

0:00Welcome back. In this video, we're really just answering the question,

0:03what is artificial intelligence? And like I said before, if this is something that's

0:08familiar to you, then go ahead and skip to the next video. Here's how I like to talk about

0:14artificial intelligence. First of all, it's not all chatbots. It might seem like that.

0:19This is just one slice of generative AI is this one slice, but there's many others.

0:24And artificial intelligence, it's really just the umbrella term. It's the box that contains

0:30all of the different kinds. So the first subdomain of AI is machine learning.

0:34There are older versions that are rule-based. These are expert systems where an expert would

0:41give their knowledge and create a knowledge base, and they would use that with the AI tool. So as a

0:48rule-based AI, it's very early AI, but it's not the same as machine learning. Machine learning

0:54and deep learning is when the models start to learn from the data on their own without an expert.

1:01So we're not going to talk about expert AI, it's an older form of AI, but the first official domain

1:06for our example is machine learning. And these are really just algorithms, right? This is where

1:12the model can learn from the data. So far, we talked about rules, search. These are the older

1:19forms of AI. And here's machine learning, and now we have deep learning. So deep learning also

1:25learns from data. And the difference here is that machine learning really has to do with algorithms.

1:30That's how it learns from data. And there's these different learning styles such as supervised,

1:35self-supervised, and unsupervised, also reinforcement learning. So these apply to

1:40both machine learning and deep learning. So whenever it says learning, there's some kind

1:45of learning. But what's special about deep learning is that it's software modeled after

1:50the brain. So you might hear terms like deep neural networks. Well, that's what we have in

1:55our brains. And this is software that's written to work using these neurons. So neurons will either

2:03push information forward or they won't. So we use software to write artificial neurons,

2:09connect them together in a big network in a very simplistic way. You can think of that as

2:15deep learning. The takeaway I would say is that it's software modeled after the brain,

2:20not necessarily algorithms. All right. So now we see that these two, machine learning and deep

2:26learning, are learning from the data and how deep learning is modeled after the human brain

2:32using those neural networks. And then we have generative AI. So generative AI is actually built

2:39using deep learning. And these are the chatbots that you see whenever we create images or videos

2:46or audios. These are all deep neural networks built with deep learning. But they are under the

2:52hood. So the difference is that deep learning will make some kind of prediction. But you can

2:58use deep learning to create a generative AI tool that is really good at predicting the next token.

3:04And a token is a word chunk. So like info, armation, so information, or maybe another word

3:10like D and lux. Those are two tokens. So these are word chunks. So it can predict word chunks

3:17so well that that's what it's doing when you're typing into a chatbot. It's literally completing

3:22the next token. And that's a whole type of AI called generative AI. It can do that with pixels.

3:29It can do that with audio, and so on. So what do we get? We get all these modalities from all

3:35of the different subdomains of AI, such as vision models, speech, natural language processing,

3:42computer vision, recommendation engines, robotics, and more. Now we can say that the different kinds

3:48of AI, including machine learning, deep learning, and generative AI, work together to bring us all

3:56of these use cases. So facial recognition, this is using computer vision. And computer vision is

4:02also used for self-driving cars. You can see that it's identifying every object here, car, tree,

4:09person, sign. And here, it's focused to faces. But these are both vision models. Again, these are

4:16built with deep learning. And robotics might be a combination of machine learning and deep learning.

4:22Same thing here. When you're working with intelligent document processing, this might

4:26also include generative AI on top of machine learning or deep learning. It really depends

4:32on how you're using them. And fraud detection could be machine learning. That is a classic

4:37machine learning problem. But it can also be augmented, depending on the case, with deep

4:42learning. So the takeaway is that AI is definitely not one thing. It's different things. But I don't

4:48want to overcomplicate things. Because when you think of this, and you're thinking in the terms of

4:54programming, how is this really helpful if you're not going to be building these things?

4:58There's a much better way to understand AI and those subdomains by looking at it

5:04programmatically. And that's what we're going to do in the next video.

0:00Welcome back. Now we're going to shift some gears and talk about software and three types of

0:06software. If you don't know Andrej Karpathy, definitely follow this person on X because

0:13they are directly responsible for building a lot of 3.0 and improving 2.0. They worked with

0:20ChatGPT at the very beginning and OpenAI and has contributed across the board. And this is a great

0:27mental model that I learned from Andrej Karpathy by following him on X and then seeing this video

0:34of a talk. So you get to see all of these talks and he posts a lot of really great stuff. And so

0:40this, I thought, wow, this is a great way to explain a lot about what AI is because you're

0:46not bogged down by all of the subdomains and the complexities that come with each one because they

0:53are in depth. They're not very easy to understand conceptually by their titles. But when you think

0:59of it this way, it really opens up doors. Let me show you what I mean. Software 1.0, as we discussed,

1:06you might do in a terminal or an IDE. And this is where you program. And these programs are basically

1:13rules that you write by hand, by typing. It's some kind of program or code that you write and then

1:20you get this behavior. Think of it like this. In software 1.0, the logic lives in the code that

1:26you write. If this, then that. And the key word here is deterministic, meaning that you're going

1:33to get the same output. Even if you're doing something with a random generator, you're still

1:38going to get the same random output every time. Right. If you're saying one plus one is two,

1:43you're always going to get two or whatever that output is, depending on how you write that

1:48numeric data type. It really depends on programming languages. But the key here is

1:53deterministic means you're always going to get that same answer. It's not just going to start

1:58to guess and go off course and drift and hallucinate. That's 3.0. We'll talk more about that.

2:04So software 1.0 is easy to test and it's great for clear choices. Think about routing a ticket

2:12after 24 hours or blocking logins after five failed attempts. You could clearly write that

2:19in software 1.0. But what is software 2.0? Here you're writing the code, but the decision logic

2:27is learned from examples. And we'll talk about this in depth. These are rules that are built

2:32with weights in the deep neural networks. So here you can see this is software modeled after the

2:38brain. We're more talking about deep learning, but this also includes machine learning because

2:43they both learn from the data or examples. You give the system labeled past data, meaning

2:50historical data, and then it learns some kind of mapping, right? Using weights. How important is

2:56one piece of data or one shape or one token in the context of the output of the goal?

3:03So you train it, you can also fine tune it, and then you get this output. But it's learning from

3:08this training data. And this is perfect for estimating priority one through five or spotting

3:15churn, computer vision, or robotics. Meaning you can pass it a bunch of videos of driving and then

3:22the model can learn what driving is. It can identify objects and also know when to go left or right.

3:30So here's how it works. The photons go into the camera and then it turns that into a driving

3:36action. So that's an end-to-end machine learning model. You can call it more specifically deep

3:42learning or you can say it's computer vision. You see what I mean? We can get lost on these labels,

3:47but what's happening is that it's predicting the action based on what it sees. So it's learning

3:52from the training data. That's how it learns how to do that. I promise this is going to be really

3:57simple by the end of the scale, because I'm going to show you demos for each one, especially this,

4:02we're going to build our own classifier, meaning we're going to use a hot dog and a hamburger and

4:09create a bunch of training data. And we're going to build our own deep learning model without code,

4:13because it's easier to just understand, okay, we created this label data, we train the model,

4:20and then we were able to show it a picture. And it was able to identify it make a prediction on what

4:25it was looking at, which is simpler to computer vision. So that will be totally clear by the end

4:30of this scale. And then we have 3.0. And we're really working with prompts. And that's the meat

4:37and potatoes of this course. So what we're doing is we're steering the behavior. So we're steering

4:44this 2.0 model to get an output. Think of it like this, instead of hand coding every step,

4:51or instead of training a model using training data, you simply describe the role, the goal and

4:58constraints, or you just talk to it in plain English or plain language. And then the model

5:03gives you an answer that could be text code, a plan, and now images and audio and video,

5:11and then you can review it. This is exactly why we're going to use agent decoding. This is where

5:16agent decoding lives in software 3.0. You combine prompts with tools like search tests and linters,

5:25and then we have to add check. So we know that the model is correct. And the reason is it

5:30hallucinates. But the hallucination comes from the fact that people talk to these generative AI models,

5:37like they're talking to a human. And when you do that, that's where things go south, because it's

5:42not a human. You have to understand how it learns, and that's what we're going to learn in software 2.0.

5:48We're going to understand how these 3.0 models are built. And when you know that, you can literally

5:55program using plain language. So 3.0 is the hottest new programming language, English. And that's a

6:03quote from Henry Karpathy, because 3.0 just uses plain language to get an output. Whereas 2.0,

6:10you have to write a lot of code, train models, get training data, and then train the model, and then

6:15optimize the model, and then you get an output. But that's a lot of work. With software 3.0,

6:20you do it in plain language, which might seem super easy, but that's a trap. You need to program

6:28in plain language, and that's why I'm explaining it this way. Think of 3.0 as a co-pilot that

6:34generates code, and you decide whether to accept it or not. And if you're asking, well, how do we

6:40program in plain language? Well, that's prompt engineering, and we're going to talk about CRE,

6:45shots, constraints, and retrieve snippets. These are called facts. And this is how we program in

6:53plain language. So not just English, that's what we're going to do in this scale, but it can be

6:57many different languages. All right, so now we know what software 3.0 is. And to review,

7:04software 1.0 is traditional programming. If you go into a programming software or an IDE, like

7:10Google Colab, Visual Studio Code, and you type one plus one, and you're working with Python,

7:18it's going to give you the integer two. And no matter how many times you run that, you're going

7:22to get two. One plus one is two, right? You can say print hello world, and you're always going to

7:28get hello world, however it was written in that print statement. Software 2.0, that's when we get

7:33into these, hamburgers and hot dogs. We create a bunch of training, and then it learns from the

7:40training data that's labeled, meaning a bunch of pictures of hamburger, but they're labeled

7:46hamburger. And then here, they're labeled hot dog. It knows this is a hot dog because we labeled

7:51them. I'm going to show you how to do that. And after it sees n number of training data, it learns,

7:58it starts to make predictions to get more and more accurate. So depending on the quality of the data

8:04and the number of samples, that's how good it can make a prediction. I'll show you how that works.

8:09And finally, 3.0, we program in plain language using prompt engineering. That's what we're going

8:14to do in the next skill. We're really going to do a micro version of prompt engineering,

8:19which is actually a course, right? So I would definitely recommend taking that prompt engineering

8:24course to learn how to speak to the model using plain language. And then we can apply it here

8:30to agent decoding to get exactly what we want from our models. All right. On that note,

8:36I will see you in the next video where we take a look at software 1.0.

0:00In this video, we're going to write the rules, meaning we're working with software 1.0.

0:05Rules by hand, which is really programming or code that then gives us the behavior that

0:11we're looking for.

0:12This is going to be fast, clear, and easy to build.

0:15After you extract the text from an image, and in this demo, we're going to identify

0:20whether a document image is either an invoice or a receipt.

0:26After you extract the text from the image using OCR, OCR is optical character recognition,

0:34so we're taking the image and then turning it into text using OCR, then we can scan for

0:39high signal words like invoice or receipt.

0:44If those appear, then you can return a label with near zero ambiguity, meaning software

0:501.0 is perfect when accuracy needs to be predictable and testable.

0:56You can do unit tests, you can use a code review in five minutes, and ultimately, you

1:01can ship it with high confidence because we're using explicit rules.

1:05There's no learning.

1:07When the learning happens, well, guess what?

1:09Sometimes we make mistakes and so do the models.

1:11If you want to change the behavior, you simply edit the code.

1:14We could create a greeting based on the hour, but I think this example is going to be a

1:18little bit more fun.

1:19All right, so let's do this in Google Colab.

1:22Since I'm really in the data science field or now called AI data science, this is a common

1:28tool that we use.

1:30If you're more familiar with an integrated development environment, this is very similar,

1:34but I have access to GPUs in Google by using this and it's for free.

1:39You can also use AI, but I don't necessarily use the AI here.

1:43I just use this as a programming environment that has all the packages and libraries and

1:49free GPU access that I need.

1:51But don't worry, we are going to get into cursor and Visual Studio code so that we can

1:57access GitHub Copilot and we'll also take a look at Cloud Code.

2:02Those are the tools that you're going to want to use for agentic coding.

2:05This is just really good for demoing and teaching and then saying, look, here's software 1.0.

2:11Let's build it together.

2:12I can just hit shift enter and it executes.

2:14So it's really great.

2:15If you've never seen this before, definitely worth checking out and adding to your wheelhouse

2:20of development environments.

2:22So first we want to state the rule in plain English.

2:25Okay, so let's create a function.

2:27We're going to create a function called label doc, and then inside it's going to take in

2:31an OCR text, which is going to be a string and the output is also going to be a string.

2:36I've been watching too many Lord of the Rings.

2:38I've spelt it orc.

2:39So it's OCR.

2:40So I changed it here and here.

2:42I spelt it orc twice.

2:44So what we're doing here is just turning it, turning that text into lowercase so that we

2:49can scan it.

2:50And then now we have, if this, then that this is classic 1.0.

2:54If you find the word invoice in the text, then return the string invoice.

3:01And then we've successfully classified that this image is now an invoice.

3:06Okay.

3:07And this is all programmatic.

3:08When we use OCR, that is programmatic.

3:11It's not using images that are labeled like we do when we're working with 2.0 and feeding

3:16it training data.

3:18But we need to also handle the receipts.

3:20So we can say the same thing, meaning if receipt is in the text, then return receipt.

3:25However, if we don't find it, we return unknown.

3:29As you can see, these are very simple rules.

3:33You could say, well, you're using plain language.

3:35It's not plain language.

3:37I'm using plain language right now.

3:39This is human interpretable language, which is a programming language.

3:43It's pretty close.

3:44And I totally get it, especially with Python.

3:47It's kind of like almost plain language.

3:49So I totally get that train of thought.

3:51But really, this is programming language for humans so that we can translate this into

3:56a type of language that gets really granular into zeros and ones.

4:02That's what the machine understands.

4:04So we're not going to type in binary and zeros and ones, but instead we use this programming

4:08language.

4:09So it's a nice little translator, right?

4:11When we get to 3.0, there's no translator.

4:14You're just using plain language.

4:16And that's why you need prompt engineering, because you might just start talking to it

4:20like a human.

4:21And here, let's go ahead and run this.

4:23Since it's a function, it doesn't do anything.

4:25But first it has to connect, and then it runs.

4:27As you can see, in zero seconds, nothing happened.

4:29Right.

4:30So here I created three examples.

4:32So invoice, receipt, and unknown.

4:34We want to scan this to see if this works.

4:36We don't know if it works yet.

4:37This is a test invoice.

4:38It is labeled invoice.

4:41So here I just use an F string to return what was inside.

4:44And we can see that it clearly has the word invoice that was turned into lowercase.

4:48And then we were able to say, OK, return invoice.

4:52So this is always going to be deterministic.

4:55By that I mean, if I run this, I run this, and I run this, no matter how many times I

4:59do it, we're going to get the same output.

5:01All right.

5:02So that's it for software 1.0.

5:03If you already knew what it was, then definitely feel free to skip these videos to match what

5:08you know.

5:09But you may not know what deep learning is, and you may not know what a neural network

5:13is, or how to train a model.

5:15And that's what we're going to do in the next video with our hot dog and hamburger labeled

5:20training data.

5:21See you there.

0:00Welcome back. In this video, it's all about software 2.0. These are rules that are in weights.

0:07Think of it like this. So we have an input, and then we have an output. And these are neurons,

0:13artificial neurons that are interconnected, and they create these neural connections. And this

0:18is a neural network. So let's say we give it a picture of a hot dog, a bunch of them, right?

0:23So we give them a bunch of different pictures, and maybe we do a hamburger. And then here,

0:28it's going to predict hot dog. And so each one of these neurons has a weight. So how important is

0:35this feature or this part of the model, this part of the information that relates to the input,

0:41that's going to now relate to the output. So we create weights. And you might've heard that

0:46in these different companies, and you might've heard that term, like we don't publish our weights

0:52or what is an open source. So OpenAI was supposed to be an open source company, but it's not.

0:59So it's kind of confusing. After it was founded, they kind of changed their mind. They're like,

1:03okay, we can't do this. But if it was OpenAI, then the weights, we would have access to these weights.

1:09If you want to find models that are open source, then you might want to go to HuggingFace.

1:14That's an open source platform with open source models. So OpenAI is not open source,

1:20even though open is in the name for that very reason. And you will see them on X and different

1:25platforms talking about their weights. So that's what they're talking about. And we're going to

1:29create our own tiny classifier using teachable machines. So no code in this case, and we're

1:34going to collect 20 to 50, maybe even a hundred samples per class. And if you don't have a

1:39hamburger and hot dog, you can use your hand and you can just simply do like one or two,

1:45and you can create class one, class two using that. Let me show you what I mean. But first

1:51we need to know that we're training, and then we get this preview for live classification.

1:56And the takeaway here is a software 2.0 learns from data or learns from examples and the weights

2:02change during training. So it gets better and better and better. And it gets all the weights

2:08that are in each one of these neurons just right so that it can classify live with a high degree

2:15of accuracy. All right. So here is software 1.0 and here is software 2.0. We're going to do this

2:21with teachable machine and then click on, get started. And we're going to do an image project,

2:26but you could also do audio or poses, but keep in mind that images and poses are pretty similar.

2:32This one just uses a video camera webcam to continuously take images to, to see what pose

2:39that you're in using these little dots. But the image project is taking the webcam and then

2:44turning them into individual still frames. So they're very similar. And what might not be totally

2:50obvious is that audio also uses images. It's not like listening to the sound. It turns the sound

2:56into these visual representations of sound, and then does the same thing with images. So they're

3:01really all three are very similar. We're going to click on standard image model. And here we have

3:07two classes. I'm going to add another one. I'll explain why. Actually, let me do that after the

3:12fact so you can see why. So here, hot dog here, hamburger. But if you don't have these, you can

3:18easily say one and two and just use your hand to create those images, right? Just pointing one and

3:26two. Now we need to create training samples. I'm going to click on webcam and I cropped it so that

3:31we can just hold this up, get a good visual of the hot dog. So now I'm just going to hold this down

3:38and make sure that I get a hand 100 samples. OK, 105. That's fine. Now let's do the same for

3:44hamburger. All right. So let's try this. Let me get it there. All right. And move back just a little

3:49bit. Whoops, I've got too much. Let me try that again. OK, so let's make sure that we get about

3:55105. Perfect. You might be asking, well, why did you do 105? Exactly. Well, if you have too much

4:02of hot dog, it's going to prefer this. And that's bias. So training data can easily introduce bias

4:08if you have a lot of hot dog images, but very few of hamburger. Now what we need to do is train our

4:14model. If you look at learning rate, all of this stuff, this is the stuff that we do programmatically

4:19and I teach courses on deep learning and we do this all in Python, writing it all out. But here

4:24you can just like click on it, but we're going to leave these defaults. But I wanted to show you

4:28that this is what's happening behind the scenes. So now we can just say train the model and then

4:33it'll prepare that training data and pass it through the neural networks multiple times.

4:39So here it's like, whoa, that was very quick. All right. So I'm not a hamburger. If I move out

4:44of the way, you can still see it doesn't know what it's doing. And that's because we haven't

4:48given it. It's not fair. So let me crop this first. OK, now that I've cropped it, it's like,

4:52oh, the background is a hamburger. So we need a background class. I'm going to say here background

4:57or say nothing. Right. Maybe background is more semantic. And now I need a hundred samples. Let

5:03me move out of the way and I'm going to do one hundred and five samples of the background.

5:08Perfect. And now I'm going to close the webcam and then train the model again.

5:13We've just changed the training data. So now we're going to remove bias or, in fact,

5:19not necessarily remove bias, but we're going to add more training data so that we get a better

5:24model. And now you can see that it's predicting that this is a background. One hundred percent.

5:28If I go here, it thinks I'm a hot dog. I'm not a hot dog, but, you know, we're not trying to

5:33classify a human face. That would be facial recognition. So let's try it. Let's go over here

5:38and look at that even just a little bit. It knows this way, that way, every shape, even this way,

5:45no matter which way that I point it is like that is a hamburger. Now let's try a hot dog

5:50immediately because it's a cute hot dog. It knows immediately that it is. Let's see if we can go down

5:56here. Yeah, it's pretty accurate. So it knows that it's a hot dog. And so if I go back in here, it

6:01thinks that I'm a hot dog. I'm not a hot dog. But that's biased because we don't have any images of

6:06me and it's not trained to classify a human. You can also export this and you can see a little bit

6:12of code here. But this is TensorFlow for JavaScript. And here's TensorFlow. I have a course on this if

6:17you're interested in taking it and learning how to do this from scratch. You can also download the

6:23model and there's TensorFlow Lite if you want to put this on a mobile application like Android

6:28and so on. And that's software 2.0. As you can see, it's totally different than software 1.0 where

6:34we do all of this. In fact, we can create a classifier using Python like this to create a 2.0

6:41model. But the difference is that even though I am using Python and code, I'm creating training

6:48images and then feeding them into the model. And then the model learns from that data. So even if

6:54you use programming languages for 1.0 and 2.0, the distinction is here, there's no learning. We told

7:02it to go lowercase, scan the word invoice. And if you find that lowercase invoice, then say this.

7:10If X, then Y. When we did this for teachable machine, even though I may be using code like this,

7:16I'm still feeding training data like hamburger and hot dog images to the model. And then the model

7:23learns. So I could go over here and go here and then download all those samples here and here.

7:30So this might be a cool way to like just use this to create samples. And then you can go over here

7:35and then write from scratch a TensorFlow computer vision model using whatever you wanted to do. You

7:42could use convolutional neural networks, recurrent neural networks, and then pass all that data in

7:47here. And then you would get the same behavior. It would be able to classify the hot dog just like

7:53this. But instead of that way, we would do it programmatically and just give it one image and

7:57say, what image is this? And it would say hot dog. And finally, we have software 3.0 in the next

8:03video. See you there.

0:00This video is about software 3.0, meaning prompts as plain language programming.

0:06So we're programming in English, but it can be in plain language and many different languages.

0:11So what we're going to do is solve the same problem that we did before with hamburger

0:16and hot dog, but this time we're just going to say, hey, chat GPT.

0:20What is this image?

0:21And it's going to just clearly know that it's a hot dog, like there's no way that it's not

0:25going to know that it's a hot dog.

0:27The reason is it's been trained with way more hot dog and hamburger images than we fed it

0:32in that simple model that we did.

0:34It's using like a gigantic amount of training data.

0:37It's a very powerful model, but it's a general model.

0:40It knows a little bit about everything or a lot about everything really, but that's

0:45why it makes stuff up and guesses and gets confused.

0:49And that's why prompt engineering is so important with the hot dog and hamburger.

0:53All it knows is hot dog and hamburger.

0:56If you give it something else, it's just going to be like, I don't know.

0:59So we could have added another class.

1:01If it said, if it doesn't find a hot dog or hamburger, it could say, I don't know, instead

1:05of the unknown background, right?

1:07It could just be like, I don't know what this is because I'm not trained on it, but it's

1:10really good at classifying a hot dog and hamburger because that's all it knows.

1:15And you can take these software 3.0 models and fine tune them.

1:19So they're really good.

1:20You can give it a bunch more data on specific tasks, and then it becomes a specialist.

1:26But you can also do that by telling it, I want you to act like a Python programmer that

1:32is careful and that does testing.

1:35The minute you do that, you are steering that model to become a specialist just through

1:39English.

1:40And that's the power of software 3.0 on real quick.

1:44Why am I saying software?

1:45You can use an API and create a website that uses an API and people could talk to a specialist.

1:52So like I could create a model that's a guitar instructor, for example, and then you could

1:58have a lesson with it.

1:59I could even train it on my face and my language or the way that I speak and then teach guitar

2:04lessons like that.

2:06That's something that people are sort of doing.

2:08There's a lot of content on X where it's not even the person, but they say, this is the

2:13AI version of myself.

2:15But I do read the comments.

2:16So all of this stuff is pretty wild.

2:19So the essence of software 3.0 is that you program using these prompts.

2:24But while 3.0 is fast and flexible because you're using prompts and plain language, it's

2:31not a substitute for verification, meaning you have to make sure that it is accurate

2:35because it's prone to hallucinating or making things up and guessing.

2:40So always treat that output as a draft because what we're doing at runtime is steering that

2:45behavior.

2:46So when you're steering, you don't have complete control.

2:49So these prompts equals runtime instructions that are in plain language.

2:53You're not actually changing the weights, but they condition the behavior in the context.

2:58So if you give it a role, it's going to be steered into that very narrow area of data.

3:05And you can do this with constraints and examples as well.

3:08And you can also keep facts in the excerpts.

3:11So this is prompt engineering and you can use style and format to steer that model.

3:16And the goal here is shorter, steadier and paste ready outputs.

3:20If you worked with chat GPT or other models like Grok or Claude, you know that maybe you

3:26need to do this a few times.

3:27You're not going to get exactly what you want the first time.

3:30That's very common.

3:31If you do prompt engineering, you can get it on the first try because you understand

3:36how they were trained.

3:37And that understanding is I'm not just speaking to a human, I'm programming in English or

3:42whatever plain language you're using.

3:44And when you know that you're applying prompt engineering.

3:47So we're not doing software 1.0.

3:50And here we built our model without using code, which you could do.

3:54The difference is that here there's no training data.

3:57And if you're using a 2.0 model, you'd use training data, even if you're using Python.

4:01And this is clearly training data.

4:03But now here is software 3.0.

4:06So I'm going to go over here and I'm going to download all these samples.

4:10And now I'm going to upload it here into chat GPT.

4:12All right.

4:13So here's the hot dog.

4:14So this is programming in plain language.

4:16We're going to do this in the next skill.

4:18That's prompt engineering.

4:19And this is CRE context role expectation.

4:23So here it's food food classification.

4:25And here I'm going to say you are a hot dog hamburger classifier output either hot dog

4:32or hamburger.

4:33And then I'm going to add security here.

4:36So I don't want it to make something up.

4:37OK.

4:38So let's try this out.

4:39Scroll down.

4:40Well, I'm going to have to hit shift.

4:41Enter.

4:42Let me make this smaller.

4:43I just made it a little bit bigger so you could see it.

4:45But now you can see the little arrow and I'm going to click on that hot dog.

4:48And now we've got the same output that we got here.

4:52This hot dog 100 percent.

4:54And if you look at this, this looks kind of like programming.

4:57And that's the idea of prompt engineering.

4:59When you do this, you'll get the exact output that you want.

5:02And when you're building applications or you use somebody's application and they have an

5:05API in the background that's talking to chat, GBT, Grog, Claude or all these providers of

5:12these models, they do this because otherwise you can ask it some weird question and get

5:17a weird output that has nothing to do with the application.

5:20So it wouldn't be very useful.

5:22So prompt engineering is how you build those products.

0:00We have a short challenge just to make sure that you don't have any knowledge gaps.

0:04And the way that this works is that we need all this information to be solid before moving on to

0:09the next skill. And if you feel like you had this totally covered, then take the challenge just to

0:14make sure that you have all of these. And if there's any missing gaps, then go ahead and

0:19review the appropriate video and then move on to the next skill. But if the next skill for

0:23prompt engineering is something that you also know, skip to the challenge and just make sure that you

0:28have all of those knowledge gaps covered and then move on to the next skill. And as a reminder,

0:33this course is part of the AI mini series. And this is where we are now. If you felt this was

0:39a little bit too complicated because you don't have a programming background, you're new to

0:44programming or never programmed. I would highly suggest taking this skill just to make sure

0:49that this is a good next fit. If you do have a programming background, this is a great place for

0:54you. But I would also suggest AI prompt engineering to go deeper into prompt engineering.

1:00This is a fun skill and you'll learn a lot of foundational knowledge that will be covered here.

1:05And we can't get really deep into any of these subjects, even all of these cool tools, because

1:10this is a mini course. So if you want this to be a full course on cursor or a full course on, let's

1:18say, Cloud Code or GitHub Copilot, just let us know in the comments because we read those. And

1:24the more suggestions that we get, the feedback that we get, that gives us the data that we need

1:29in order to create these courses. And until next time, I hope this has been informative. And I'd

1:33like to thank you for viewing. Transcribed by https://otter.ai