Lesson video

Hello, my name is Mrs. Holborow, and welcome to Computing.

I'm so pleased you've decided to join me for the lesson today.

In today's lesson, you'll be exploring predictive and generative AI systems. What's the difference between the two and when should each type be used? Welcome to today's lesson from the unit, data science, AI, and machine learning.

This lesson is called predictive versus generative AI.

And by the end of today's lesson, you'll be able to explain the differences between predictive and generative AI systems. Shall we make a start? We will be exploring these key words in today's lesson.

Let's take a look at them together now.

Predictive AI.

Predictive AI, machine learning algorithms use data to identify patterns and make guesses about the future.

Generative AI.

Generative AI, a type of artificial intelligence AI designed to generate content such as text, images or sound.

Look out for these key words throughout today's lesson.

Today's lesson is broken down into two parts.

We'll start by comparing predictive and generative AI, and then we'll move on to discuss suitability of approaches.

Let's make a start by comparing predictive and generative AI.

Sam has a question before we start.

"What's training data?" Can you remember what training data is in AI systems? Maybe pause the video whilst you have a think.

Training data is a set of data and examples processed by AI systems so that patterns and relationships can be identified.

All AI systems rely on training data to perform tasks by using patterns identified in the training data.

Two common types of artificial intelligence are generative AI systems, predictive AI systems. Note that all AI systems are designed and developed by humans to complete useful tasks.

Izzy's got a really good question.

"Do generative and predictive AI systems have different uses?" What do you think? Maybe pause the video and have a quick think.

Generative AI technology is used to generate new content such as images, text, computer code, music and videos.

Predictive AI technology is used to make guesses about the future based on patterns identified in training data.

So they do have different uses.

Okay, I have a question to check your understanding.

What is the main purpose of a generative AI system? Is it A, to identify patterns in training data without producing new content? B, to generate new content such as images, text, or music? Or C, to predict future weather forecasts using training data? Pause the video whilst you think carefully about your answer.

Did you select B? Well done.

The main purpose of generative AI system is to generate new content.

Generative AI applications are built to generate creative content such as sound, images and text.

As an example, Google's Assisted Melody uses AI to harmonise your music in the style of a composer of your choice.

Predictive AI technology uses data to identify patterns and make guesses about the future.

Common applications of predictive AI include spam filtering for emails, weather forecasting, and purchasing recommendations in online stores.

Okay, time to check your understanding.

State which description matches the approach.

So the two approaches are predictive AI and generative AI.

And the descriptions are used to generate new content such as images, text, computer code, music, and videos.

Used to make guesses about the future based on patterns identified in training data.

Pause the video whilst you have a go.

Did you get them right? Well done.

Generative AI is used to generate new content such as images, text, computer code, music, and videos, whereas predictive AI is used to make guesses about the future based on patterns identified in training data.

Sam says, "Predictive and generative AI technology both seem the same to me.

How are they different?" That's a really good question, Sam.

Izzy says, "We can compare some of the features of each.

That will help us to understand the differences.

That's a good idea, Izzy.

One difference between generative and predictive AI technology is that they're designed to produce different types of outputs.

So generative AI technology is used to generate brand new content such as images, stories, or music.

Predictive AI technology analyses data and tries to predict what might happen next.

So the prediction is the outcome.

Generative AI systems are used to generate new things like text pictures or sounds.

So this image of the cat has been generated using an AI image generator.

It generates new content based on patterns it has found in large amounts of training data.

Predictive AI systems are used to estimate what is likely to happen based on past data.

It makes predictions.

For example, whether someone might miss a payment on a credit card repayment, or which advert a person might most likely click on next.

Sam says, "So generative AI technology generates new content and predictive AI technology estimates what is likely to happen based on past data?" Izzy says, "Yes, that's right.

But sometimes there isn't a clear line between the two types of system because they can be used together to complete tasks." Generative and predictive AI systems can be combined in applications.

For example, a virtual assistant might use predictive AI systems to guess what help is needed from a user prompt.

Then it might use generative AI systems to produce and output a written explanation.

Image generators combine generative and predictive AI systems to produce images step by step.

Predictive AI systems estimate what image features should come next.

Then generative AI systems use these predictions to generate new images based on patterns in the training data.

Sam says, "Both predictive and generative AI systems can make errors." You're right, Sam, they can.

Because predictive and generative AI systems can make mistakes, it's important to check their output carefully.

Izzy says, "Generative AI systems do not always produce accurate results and can sometimes create information that sounds real but isn't actually true." It's important to fact-check the output of generative AI systems. Predictive AI also doesn't give correct answers.

It can make inaccurate predictions if the training data is biassed or incomplete.

Always check and question the results of predictive AI systems before making decisions based on what they output.

Sam has a good question here.

"Do developers know why AI systems generate or predict certain outputs?" Maybe pause the video whilst you think about Sam's question.

Izzy has an answer.

"Even developers can't easily explain exactly how AI systems produce certain outputs." This is because the internal processes are based on complex patterns in training data rather than clear rules.

So there are lots of differences between predictive and generative AI, but they also share some similar features.

Izzy says, "I think a table might be useful to help us compare them." Okay, we are moving on to our first task of today's lesson, task A, where we are going to compare predictive and generative AI.

Complete the table to help outline some features of predictive and generative AI systems. So the features we have are the main purpose, how it works, and where it is used, so used in.

And then we've got the two columns for the data for generative AI and predictive AI systems. Pause the video whilst you complete the table.

How did you get on? Did you manage to complete the table? Well done.

Let's have a look, a sample answer together.

So for generative AI, the main purpose is to generate new content.

It works using identified patterns to generate new content.

It's used in chatbot applications, writing and design tools and image generators.

Predictive AI systems main purpose is to predict what might happen.

They work by identifying patterns and making guesses based on past examples.

They're used in business or weather forecasting, medical predictions, and spam filters.

Did you have some similar response? Remember, if you need to pause the video and add any detail to your answer, you can do that now.

For part two, in your own words, explain the difference between predictive and generative AI models.

Pause the video whilst you have a go at the question.

How did you get on? Did you manage to explain the difference between predictive and generative AI? Let's have a look at a sample answer together.

Generative and predictive AI are both types of AI that use training data to identify patterns, but they are useful in different ways.

Generative AI systems are used to generate new content such as images, text, and music.

Whereas predictive AI systems estimate what might happen in the future, and are used for things like forecasting weather and making recommendations.

Both types of AI systems can produce useful results, but they can also make errors.

It is important to carefully check outputs from both systems. Okay, we're moving on to the second part of today's lesson where we're going to discuss suitability of approaches.

Image generators use generative AI systems to produce new images from text prompts.

They generate new visual content because the systems have been trained to identify patterns from millions of images in the training data.

Generative AI systems can be used to output new video content by inputting text, images or audio.

Generative AI systems generate videos by identifying patterns from millions of existing videos in training data.

Code generators can help programmers by generating software code from simple descriptions in a user text prompt.

The generative AI systems have been trained to identify patterns in millions of code examples in the training data.

So you can see here there's a bit of a pattern there, all relying on this training data.

Large language models, or LLMs, are a form of generative AI technology and can be used to generate text like stories, summaries and translations.

Chatbot applications use LLMs to generate text responses based on patterns in training data.

Okay, time to check your understanding.

I have a true or false statement for you.

Only generative AI systems require training data to complete tasks.

Is this true or false? Pause the video whilst you think about your answer.

Did you select false? Well done.

All AI systems rely on training data to perform tasks by patterns identified in the data.

Spam filters are designed to detect and block unwanted or malicious emails or messages before they reach a user's inbox.

They use predictive AI systems that can identify patterns from training data to classify emails as wanted or unwanted.

Weather apps are designed to provide users with current weather conditions and forecasts of what the weather might be like days or weeks in advance.

Whether apps use predictive AI systems that have been trained to identify patterns in large amounts of past weather data.

Online stores can make recommendations and suggest products you might like to buy.

They use predictive AI systems that can match your browsing history with patterns identified from other customers purchasing behaviours in the trading data.

Okay, another true or false statement for you.

The output of generative and predictive AI systems is always accurate and based on facts.

Is this true or false? Pause the video whilst you have a think.

That's right, it's false.

Generative and predictive AI systems can produce convincing output, but it may not always be accurate or based on facts.

Fact checking is always important because AI systems generate output based on patterns in the training data, not verified information.

Okay, we're moving on to our second task of today's lesson, task B.

And you've done a fantastic job to get this far, so well done.

For each of the tasks below state whether predictive or generative AI should be used, and discuss the suitability of the approach you have selected.

So the first example has been done for you.

The task was to produce an image to advertise a school disco.

Generative AI has been chosen and the explanation for suitability is a text prompt could be used to generate a new image.

The three other tasks you have got to fill in are produce some Python code to create new student IDs, produce suggestions for a library user's next read, produce a chatbot application that answers students' new questions.

Pause the video here whilst you complete the table.

How did you get on? Did you manage to suggest whether predictive or generative AI would be most suitable? Well done.

Let's have a look at sample answer together.

So produce some Python code to create new student IDs is generative AI.

A text prompt could be used to generate python code.

The user could then refine the code and make sure it works correctly.

Produce suggestions for a library user's next read is predictive AI.

Predictive AI can be used to find similarities and patterns in library rentals and makes suggestions.

Produce a chatbot application that answers students' new questions.

This is generative AI.

An LLM model can generate responses based on training data.

Remember, if you want to pause the video and make any corrections to your table, you can do that now.

For part two, explain why generative AI would not be suitable to use for a weather forecasting app.

Pause the video whilst you answer the question.

How did you get on? Let's have a look at a sample answer together.

You were asked to explain why generative AI would not be suitable to use for a weather forecasting app.

Generative AI is used for producing new content.

Predictive AI is used to make predictions.

Although both rely on training data, a weather forecasting app would not be a suitable task for generative AI.

Predictive AI would be more suitable for the app as the system would need to make predictions based on previous data.

Okay, we've come to the end of today's lesson, predictive versus generative AI.

And you've done a fantastic job, so well done.

Let's summarise what we've learned together in this lesson.

Predictive AI makes predictions using patterns found in existing data.

Generative AI is a type of AI that creates new, original content based on patterns it has learned.

Predictive AI is typically applied in tasks like forecasting future trends and classifying data into categories.

Generative AI can be used to produce a wide variety of new outputs, including text, images and code.

I hope you've enjoyed today's lesson, and I hope you'll join me again soon.

Bye.