Chat GPT is a powerful tool for generating text using artificial intelligence, but when it comes to analyzing text, using a model focused on understanding can produce more consistent and accurate results. Natural Language Understanding (NLU) is a type of AI designed specifically for analyzing and understanding text data. In contrast, GPT (short for Generative Pre-trained Transformer) is primarily designed for generating text, rather than understanding it. To accurately and reliably analyze text, it's advisable to use a tool specifically dedicated to NLU.

{{cta}}

In order to demonstrate this, in this post I’d like to share how I used Chat GPT to write my spec script, and OneAI to understand it.

Lets dive in:

Step 1

Ask GPT-3 to generate text, here we asked him to create an episode of a kids cartoon show. Here’s the prompt we used: 

“write a funny episode of SpongeToad where Squidword starts a successful NFT business”

Step 2

In order to add some color and provide a visual representation to your script, you can embed images from DALL-E (yes, this is also a generative model).

And here is the result:

Squidword has always been something of an art aficionado, and his latest venture is no exception. He's decided to start an NFT (non-fungible token) business, selling digital artworks that are authenticated and verified on the blockchain.

To everyone's surprise, Squidword's NFT business takes off. He's able to create beautiful and unique artworks that people can't get enough of, and he's even able to hire a few employees to help him keep up with demand.

But as Squidword's business starts to flourish, SpongeToad and Batrick can't help but feel a little left out. They decide to try their hand at creating NFTs of their own to help Squidword, but their artworks are, shall we say, less than successful.

Despite their best efforts, SpongeToad and Batrick's NFTs are met with little enthusiasm. But they're not about to give up that easily. Determined to make it in the world of NFTs, they come up with a plan to help Squidword's business.

They start by creating fake accounts and leaving positive reviews on Squidword's website. They also hack into his computer and change the prices of his artworks to ridiculous discounts.

At first, Squidword is grateful for the help. His business starts to thrive, and he's able to hire even more employees. But as SpongeToad and Batrick's efforts to help Squidword's business continue, things start to get out of hand. They start posting bizarre and spammy messages on social media, trying to drum up more business for Squidword. They even go as far as creating fake artworks and passing them off as Squidword's own.

As you might expect, this doesn't go over well with Squidword's customers. They start to feel overwhelmed by the constant barrage of messages and offers, and many of them start to tune out.‍

To make matters worse, SpongeToad and Batrick's meddling starts to attract the attention of the authorities. They're accused of fraud and deception, and soon Squidword's business is under investigation.

In the end, Squidwordis forced to shut down his NFT business and go back to the drawing board. As he laments his failure, SpongeToad and Batrick can't help but feel a little bit guilty. "We're sorry, Squidword," SpongeToad says sheepishly. "We didn't mean for things to turn out like this."

But Squidword is not one to hold a grudge. "It's okay, SpongeBob," he says with a sigh. "I guess I'll just have to come up with a new business idea. I'm sure I'll think of something."

As Squidword plots his next move, SpongeToad and Batrick can't help but feel a little bit relieved. "We'll be sure to stay out of your way this time," they promise. "We promise not to try and help again."

despite their good intentions, SpongeToad and Batrick's efforts to help Squidword's NFT business have resulted in its downfall. As they try to make amends, they can't help but feel guilty and ashamed.

But they're not the only ones feeling the effects of the failed business. Mr. Krabs, who had invested heavily in Squidword's NFTs, is left with nothing to show for it. He's devastated by the loss, and he can't help but blame SpongeToad and Batrick for their role in the disaster.

As the dust settles, Squidword decides to take a break from the world of business and focus on his true passion: music. He spends his days practicing his clarinet and trying to forget about the failed NFT venture.

But he can't escape the guilt and regret that linger in the back of his mind. He can't help but wonder what might have been if he had just been able to make a success of his NFT business.

In the end, SpongeToad and Batrick are left to pick up the pieces and try to make amends for their mistakes. They vow to be more careful in the future and to think more carefully about the consequences of their actions.

As they sit on the shores of Bikini Bottom, watching the sunset, they can't help but feel a sense of sadness and regret. But they know that they must move on and learn from their mistakes, so that they can be better friends to Squidword and Mr. Kraps in the future.

Step 3

Now it’s time to confirm that our talented writer (Chat GPT) has covered all the necessary topics and to prepare the story for publication.

We will need to extract the topics, generate a summary, create a table of contents, and come up with a catchy headline. These steps will help to organize and present the content in a clear and effective way, making it easier for readers to find, understand and engage with the story. After completing these tasks, we will be ready to publish the story and share it with our audience.

Process the text generated in step 1 using OneAI. Here is the output:

We received the topics of the script, a couple of headlines, chapters, emotions, and also the highlights and the summary.

Check out the code we used:

```

import oneai

oneai.api_key = "<YOUR API KEY>"

pipeline = oneai.Pipeline(

 steps = [

oneai.skills.Highlights(),

oneai.skills.Headline(),

oneai.skills.Topics(),

oneai.skills.SplitByTopic(amount="more",),

oneai.skills.Emotions(),

oneai.skills.Summarize(

  auto_length=False,

  find_origins=True,

  max_length=200,

  min_length=100,

),

 ]

)

output = pipeline.run(text)

```

AI-powered content machine for marketing managers?

Absolutely, here’s how:

1. Provide briefs to Chat GPT and ask it to write the text you require

2. Analyze the text with OneAI in order to:

• Ensure that the generative model covered all the topics you asked it to
• Create and test different headlines
• Improve content consumption with chapters, highlights, and table of contents
• Enhance searchability and discoverability with auto-tagging
• Prepare a summary for your cover page or newsletter
• Generate quick highlights that can be shared on social media
• Take it to the next level - go ahead and monitor talkbacks and comments to prevent abusive language, bullying and harassment

Conclusion

The combination of generative model with NLU capabilities is a powerful tool for marketing teams, content writers, and community managers.

‍