Categories
SEO Software Tools

4 Ways to Use AI Now in Your Marketing Program


In my last article, I saw artificial intelligence getting better and better at answering our questions, regardless of subject or sector.

Using GPT-3 technology, we have demonstrated that AI can pass SEO multiple-choice tests and solve SEO case studies.

Other technologies, such as Deepmind’s Gopher, are superior to GPT-3 in the humanities, social sciences, medicine, science, and mathematics.

The following graph shows the accuracy of the answers provided by Gopher, UnifiedQA, GPT-3, and human experts.

Depending on the theme, there is a narrow gap between the level of AI and the level of professionals.

This also suggests that AI can exceed the level of non-experts.

Performance of large-scale multitasking language understanding (MMLU) benchmarks categorizedImage from Deepmind, June 2022

Today, AI can be a useful support resource for many marketing topics.

Let’s see how to work with AI and, more importantly, how to integrate AI into your business.

4 ways to interact with AI tools

Focus on the main ways of interaction to successfully set up the best AI-human combination.

1. Using a web browser

The most interesting kind of dialogue is connecting AI to a web browser using a back office or Chrome plugin.

There are many potential applications because it can assist users in advanced projects such as document classification, writing assistance, meta tag generation, text extraction, and even suggesting new topics.

You can connect the AI ​​to the tool via a simple Javascript call that asks you to perform a specific task.

The following is an example of JS and GPT-3 integration.

This example is oversimplified and shows that with 20 lines of code and appropriate instructions, language models such as GPT-3 can be easily connected.


var xhr = new XMLHttpRequest();
xhr.open("POST", 'https://api.openai.com/v1/engines/text-davinci-002/completions');

xhr.setRequestHeader("Content-Type", "application/json");
xhr.setRequestHeader("Authorization", "Bearer sk-RkXXXXXXXXXXXXXXXXXXXXXXXX");

xhr.onreadystatechange = function () {
if (xhr.readyState === 4) {

if (xhr.status==200){
var data=xhr.responseText;
var jsonResponse = JSON.parse(data);
answerGPT3 = jsonResponse["choices"][0]['text'];
console.log(answerGPT3);
}
else {
console.log('API error');
console.log(xhr.responseText);
}
}};

var prompt = "List 50 concepts about … "

var data = `{
"prompt": "PROMPT",
"temperature": 0,
"max_tokens": 256,
"top_p": 1,
"frequency_penalty": 0,
"presence_penalty": 0
}`;

data = data.replace('PROMPT', prompt)

xhr.send(data);

If your back office supports Javascript and you have experienced developers, integrating advanced AI features will never be easier.

To test this initial connection, create a Chrome plugin to evaluate AI for its ability to respond correctly to certifications designed for professionals.

To do this, use the Chrome plugin that can capture all kinds of text with optical character recognition.

Then make the page as easy to understand as possible by leveraging another Chrome plugin that modifies the CSS of the page.

In the exercise example, each answer in the multiple-choice test was assigned to the corresponding letter.

Then, in the previous 20 lines of code, I sent the AI ​​instructions to generate a result in the text field.

With this program called “Asimov’s Test”, AI was able to pass several certifications.

Next, I tested AI on a medical theme and achieved a score of over 60% without prior training in a particular area.

This confirms that by choosing the right subject, the results generated by AI can be very helpful to the team in improving their day-to-day operations.

SEO rating from LinkedInImage from LinkedIn, June 2022

2. Using data visualization tools

In the last few months, tools have begun to emerge that make it easier to generate documentation and write code.

One notable use case is simply to generate an instructed dashboard or SEO tool.

Currently, we have the following open source tools Streamlit There are very advanced components of data visualization or data manipulation.

By providing the appropriate steps, you can easily request the generation of an app that interacts directly with your data.

For example, you can generate a web application with a complete interface and feature code.

This method is very recent because it uses a language model that uses only computer code. And again, the results are very impressive.

In the graph below you can see all the most common code generators and the data they were created with.

  • CodeParrot: 50GB.
  • GPT-3 Codex: 159GB.
  • InCoder: 216GB.
  • PolyCoder: 249GB.
  • AlphaCode: 715.1GB.
  • CodeGen: 1.38TB.

It is possible to generate applications in many languages. The main ones are Java, C, JS, and PHP.

The most popular code generatorImage via Huggingface.co, June 2022

Whether you are a beginner or an expert, we encourage you to give it a try. AI can create SQL queries and graphs in seconds.

Practical use is important if you want to use your favorite tools well.

3. Use of chatbots

You can also answer very specific questions by creating a chatbot to clearly specify the role that the instructions should play.

Here, we ask the chatbot to respond like a doctor, with kindness and humor.

Chatbot answersScreenshot by author, June 2022

Therefore, AI-based chatbots can provide personalized advice and recommendations based on customer preferences. Don’t hesitate to adjust your AI to respond in a particular way.

A good example is Danny Richman, who created an AI version of Google’s John Mueller called MuellerBot.

Based on the above principles, this bot answers as if John Mueller himself was answering SEO questions.

The answer can be very accurate, so it’s fun and anxious.

MuellerBotImage from Danny Richman, June 2022

4. Using the AI ​​Assistant Program

Finally, SEO’s AI Assistant is a program that runs in the background and applies SEO fixes in the event of poorly configured pages or traditional errors.

The first such application dates back to 2016 when Facebook was Implementation of automatic bug fix using Getafix..

The assistant prepares a fix template that is applied and reviewed by humans before being rolled out, based on all the bugs fixed in the past.

This is very true for SEO where meta tag title, description, pagination and linking concerns are known to be typical issues.

Facebook codebaseImage from Engineering.fb.com, June 2022

To do this, use GPT-3 in edit mode and use the appropriate steps to modify your SEO page.

Below are my instructions:

  • Add a title with the H1 tag at the beginning of the text.
  • To the most important words in the body of the textAdd a link.
    • Use to create a useful outlink at the end of the text.
    • Add a YouTube video to the body of the text.
    • The top five concepts are shown in bold.
    Change seo page with AIScreenshot of OpenAI, June 2022

    Examining the generated text reveals that the results are excellent. The H1 title summarizes the article, the bold words are accurate, and the YouTube video and outbound links are related to the theme.

    Text generated using AIScreenshot of OpenAI, June 2022

    In short, using AI Assistant can save you a lot of time.

    Note: All links are dummy links, but you can connect them all to a linked database and use the mapping table to replace LINK1 with a link in the database or CSV file.

    Now you understand the potential for automating these types of tasks.

    Now that you know the different ways to interface your language model with existing tools, don’t hesitate to implement the methods that work best for you:

    • Chrome plugin.
    • Directly to the CMS.
    • Via chatbot.
    • Data visualization.
    • AI assistant.

    Other resources:


    Featured Image: Graphic Farm / Shutterstock





Standing Desk Reviews