What Is Generative AI? (With Examples)

Generative AI has received a lot of attention as of late, and it’s really no surprise. There are a handful of significant models — ChatGPT and DALL·E, to name a few — that seem to be game-changers in this space. 

But what exactly is generative AI? In our comprehensive guide, we’ll break down what it is, how it works, and the best examples of AI models you can start using today.

What Is Generative AI?

Generative artificial intelligence (AI) is a type of AI that can provide users with human-like output to various questions or prompts. These types of content can include audio, video, text, and imagery, among other things.

Generative AI is a bit different from traditional AI systems. Traditional AI entities use patterns to make educated predictions. On the other hand, generative AI takes it a step further through unique content creation.

History of Generative AI

Although generative AI might seem like the hot new thing, it’s actually existed for a while. From Georges Artsrouni’s 1932 creation of his “mechanical brain” to Google’s 2023 plans to release its model Bard, there’s much to explore in this space.

Though it’s not a complete list, here’s a brief history of some of the most important discoveries and milestones in the lifespan of generative AI:

What’s the Difference Between AI and Machine Learning (ML)?

AI and machine learning (ML) are often confused or used interchangeably. While AI is the overarching category — involving machines completing tasks with human-like output — machine learning is a simple type of AI. 

Machine learning refers to AI models that can “learn” — without any outside human direction — from data patterns to improve. 

How Does Generative AI Work?

There are different generative AI models. One such example is text generation. These models use natural language processing methods (NLPs) to take characters — such as letters and punctuation — and translate them into sentences. 

Generative AI uses a kind of deep learning to create new content called generative adversarial networks (GANs). Two neutral networks are responsible for making up a GAN: a generator that produces new information and a discriminator that evaluates that info. The generator and discriminator work in tandem: the generator enhances its outputs depending on the feedback it gets from the discriminator until it generates content that is indistinguishable from real data.

That being said, these models all rely on AI algorithms to work and learn from the data they’re built on. Things like natural language processing techniques can also house racism, sexism, homophobia and other biases within the training data. 

People who create these models will then use a neutral network that lets the model read and respond to prompts or questions through generated text. For example, GANs, as well as variational autoencoders (VAEs), are often used. 

What Is Generative AI Used For?

Generative AI can be used for lots of purposes. These include content generation like video, audio, images, and text, among other things. 

For example, imagery can be created based on images that already exist. Certain mobile apps like YouCam Makeup and FaceApp use an existing image — let’s say, a selfie you took — and transform it to make you look 50 years older. Apps like Dream by Wombo can generate a piece of art based on a prompt. When we inputted a prompt, “a dog’s dream,” the Dream app generated this image for us:

In terms of text, people can use generative AI to write poetry, scripts, and news articles. In a similar vein, this type of AI can also be used to create sound effects and music tracks.

Examples of Generative AI

Because the industry of AI has taken off in recent years, there are plenty of examples of generative AI. We’ll discuss three examples that have been making headlines lately: ChatGPT, Dall-E, and Bard. 

ChatGPT

ChatGPT has gotten a lot of buzz lately. It’s a chatbot that’s powered by AI. As mentioned above, it uses NLP to function. 

This model has been developed using tons of information from a plethora of different places, such as websites and books. Because it’s been trained on this data, it can successfully produce language and communication capabilities that are human-like in nature. ChatGPT can help people in a number of ways, such as: 

• Answering questions
• Offering recommendations
• Participating in everyday conversation
• Producing content

For example, we asked ChatGPT to write a poem about bees for us, and this is what it produced: 

"The bees buzz and hum,

A symphony of nature’s drum.

Dancing from flower to flower,

They work with all their power.

Golden pollen clings to their legs,

For nourishment for her hive,

The lifeblood of her tribe.

Without the bees, what would we do?

No fruits, no nuts, no honey, too.

Their importance cannot be ignored,

Their work, a gift to be adored.

So let us thank the bees each day,

For all they do, in their own way.

The buzzing, busy, beautiful bee,

A symbol of nature’s harmony

A gift to the queen bee, who begs."

For more information, check out this explainer on ChatGPT and how it’s used: 

DALL·E (and DALL·E 2)

DALL·E is another great example. It’s another version of GPT-3 — a transformer language model — that’s designed to use text descriptions to produce images. 

Here’s what happened when we asked it to create images of a bust of Cleopatra:

Here’s another example. We gave DALL·E some more creative liberty and asked it to generate images of a “digital art of a tiny bug on a monstera leaf.” Here’s what DALL·E came up with: 

If you want to learn more about DALL·E and DALL·E 2, check out this video explainer:

Bard

Yet another promising example of generative AI is Bard. Bard, created by Google, is a version of language models for dialogue applications (LMDA) that’s optimized and trained on data from sources open to the public. 

Bard is able to recognize language patterns and use them for predictions. One of the examples Google gives in its explainer brief is that as the model learns, it would be able to predict that the following word in the phrase “peanut butter and …” is more likely to be “jelly” than an unrelated word like “shoelace.” 

Described as a “creative and helpful collaborator,” it can complete more creative tasks like outlining blog posts, creating packing lists for your next trip, and even party planning. 

If you’d like to try out Bard yourself, you’ll have to add yourself to the official waitlist. Once you’ve been added, you should eventually receive an email invitation to try it out for yourself. 

You can learn how to use Bard here:

Other Major Players in the Generative AI Space

The generative AI space is becoming more competitive by the day. However, generative AI models are expensive to create. Generally, that means the major players in this space are established brands and companies. 

For example, the main brands behind generative AI research and models include:

• Google
• Meta Platforms
• Microsoft
• IBM
• OpenAI

In addition to creating their own respective generative AI models, these companies are also funneling money into both development and research in this space. 

China also has some pretty significant players in generative AI, including: 

• Alibaba
• Tencent
• Baidu

What Can Generative AI Be Used for?

Essentially, the uses for generative AI are endless. Its impact on various businesses and industries is already significant, and researchers believe its usefulness is only in its beginning stages. 

For example, some solutions solved by generative AI include things like: 

• Locating bugs in existing code
• Creating software
• Conducting analyses on large datasets
• Generate different versions of marketing copy
• Create job descriptions
• Provide human-like SMS assistance and customer support (without wait time)
• Summarize complex text into plain language
• Conduct data entry tasks

What do these examples translate to? In a nutshell, they demonstrate this type of AI’s ability to impact businesses by improving a consumer’s experience, expanding business models, and increasing productivity among the workforce, among other benefits. 

Your imagination is the limit in terms of the solutions generative AI offers and will continue to provide.

How Generative AI Is Transforming Public Speaking Coaching

Another area in which generative AI is transforming the industry is in public speaking coaching. Take Yoodli, for example. 

Yoodli is a speech coach that’s powered by AI. It generates human-like feedback on speeches, such as suggestions on how to:

• Rephrase parts of your speech to improve flow and cohesiveness 
• Be more concise in certain areas
• Add additional information to improve the overall speech

In addition to providing helpful metrics, this AI coach will offer specific coaching comments that are unique to your speech and speaking patterns. 

Hands down, it’s one of the best personalized, private, judgment-free communication coaches out there. 

What’s Next?

Generative AI is changing the world, and as an industry, it’s in its beginning stages. 

As Bill Gates said, it’s “so incredible, it will change society in some very deep ways.” And to his point, it already has.

However, Stephen Hawking offered humanity a warning about artificial intelligence. “It will either be the best thing that’s ever happened to us, or it will be the worst thing,” he said. “If we’re not careful, it very well may be the last thing.”