I use AI for blogging and I made $115,000 last year from my website. Here are the exact chatbots I use and how I get them to write in my voice.

• Samantha North, a blogger who runs two websites, started experimenting with AI tools last year.
• She found they helped her generate ideas related to topics and write elements of her blogs.
• She shared the prompts she used to train AI tools to help produce her blogs.

This as-told-to essay is based on a conversation with Samantha North, a Portuguese blogger and content writer. Insider has confirmed her job and income. The following has been condensed and edited for length and clarity.

I maintain two websites. In September 2020, I launched Digital Émigré to provide resources for people who want to relocate to European countries with the long-term goal of obtaining second citizenship there.

I got the idea when I decided to move to Portugal from the UK after the UK left the EU.

My personal website, samanthanorth.com, is where I teach people how to make money through blogging, search engine optimization, LinkedIn, and other forms of digital marketing.

My blogs earned me $115,000 last year.

I make use of SEO keyword tools.

I have experience in journalism and content creation. I used to write my own blog posts before I started using AI.

I would begin by reading articles online and conducting research on the subject. I’d also consider what people might be looking for based on my personal experience of moving countries.

I’d confirm this by using the SEO-keyword research tools Semrush and Ahrefs to see how the keywords performed.

These tools can tell you how much competition a keyword topic has from other sites.

Then I’d put together a blog outline. Once I had that, I’d write the various sections of the piece using the research I’d done.

At first, I didn’t like AI-generated content.

I’ve always been a terrible manual typist, so I began using AI for voice typing.

Last year, I began experimenting with an artificial intelligence (AI) tool called Jasper for marketing content. I’d type in simple prompts like “write a blog post about” a topic. At first, it felt like magic. But I grew accustomed to its style and tone.

AI is fantastic for challenging my own ideas. When I’m considering a new direction for a blog, I can ask AI if it fits well with the existing content, and it will provide a well-rounded and useful response. However, I cannot simply insert AI content into my blogs because it would not sound like me. It’s far too bland.

As Jasper focuses more on marketing content, I began using ChatGPT this year. It was more adaptable and versatile for me.

I signed up for ChatGPT Plus, which costs $20 per month.

I use AI to generate ideas

ChatGPT has not been able to replicate my keyword-research process. It can’t generate the exact keywords I’m looking for, with their volumes and levels of competition — at least, not yet.

However, it is useful for “topic clustering” — generating ideas about or related to a topic. For example, if I wanted to use AI to generate blog ideas about Portugal, I’d create a prompt in which I asked it to generate a list of semantically related keywords.

They may not be the exact keywords that people are looking for. However, seeing them generated helps to open my mind to possibilities I might not have considered. Then I can begin plugging those ideas into the keyword-research tool in order to find one of the specific keywords that people are looking for.

I’ve started using Claude 2 in the last few months.

The interface is easier to use than ChatGPT’s. The text it generates is much more similar to what I would write. It’s straightforward and to the point. I don’t need to be as involved as I am with ChatGPT.

Personal experience and opinions, in my opinion, are still very important in posts. I believe Google’s algorithm values experience, expertise, authority, and trustworthiness.

I use AI to outline the blog, as well as for subheadings, post titles, and meta descriptions.

I’ve recently begun using ChatGPT to create introductions for blog posts. “Create a compelling hook for a post about X, Y, or Z, and use the copywriting framework PAS — which stands for problem, action, solution,” for example.

Once I’ve written a post, I’ll enter it into ChatGPT and instruct it to write a summary of the piece as well as some actionable steps for the reader in the article’s body.

I’ve trained ChatGPT to write in my voice

ChatGPT was trained using my previous blog posts.

If there is a generic section of a blog, such as a history of Portugal, I would probably have ChatGPT generate it. Following that, I’d edit it to ensure the facts were correct and add external links.

If it generates a section that’s flowery and full of adjectives, I tell it to “rewrite the text in a less flowery and more down-to-earth way.” That usually results in something close to what I’m looking for.

I practice my voice by feeding it three posts and asking it to describe the tone, voice, and style to me.

Then I instruct it to generate text in that style. It may require a few adjustments, but I’ve found that it works well.

I still write a lot of my own posts.

While using AI has allowed me to produce more content, it still takes a significant amount of time to manually edit, regenerate, and fact-check it.

I don’t rely on AI to do everything for me. I don’t believe AI-generated content is as valuable to the reader as AI-generated content edited by a human. It is deficient in human expertise and experience. I’m concerned that the internet will become overrun with generic, AI-generated content. But I use it sparingly because I believe it produces the best results.

If I’m writing something that requires a lot of personal expertise, or if I’m writing about something specific or detailed, I’ll still do my own research.

It’s about striking a balance between using these tools and weaving human experiences and data together.