I wanted to find out if my blog posts are “indexed” in ChatGPT and other tools.
To test this, I asked a very specific question.
“How can I publish from Obsidian to WordPress”.
I wrote a tutorial on that in February 2023, and this one is getting some traffic from Google.
Now, I wanted to check if the information from my how-to is also the basis for answering the question inside AI chat tools.
I asked in
- Perplexity.ai
- Google Bard
- Claude
- ChatGPT
Here are the results:
Perplexity.AI
Perplexity AI lists Fresh van Root as the first source for this question. It is also the AI chat app with a user experience that invites you to explore external sources on the topic.
It prominently features its sources on top of the provided answer, and in addition also surfaces additional media on the right-hand side.
Google Bard
Google Bard is also referencing Fresh van Root as the first result for this how-to-question. I would have been suprised if the blog post would not have been included in the sources list, as it is indexed in Google search index.
ChatGPT
ChatGPT answers the question, but GPT 3.5 and 4.0 do not provide the source. The reason for that is simple: the data cutoff date was January 2022, meaning that anything published after that date was not used to train the language model.
Claude.AI
Claude is also not providing sources and like ChatGPT saying “I used my general knowledge of Obsidian and WordPress as commonly used software tools for writing and publishing content online respectively.”
What are my first thoughts on this?
It is interesting to see that my content is used by AI platforms to answer a specific question.
I would of course like for all AI systems to prominently feature its sources by default, like Perplexity AI is doing it.
Google Search Console gives me insights into how my domain freshvanroot.com performs in Google search. What could such a tool look like for AI chat systems? As a blogger, I would be interested if/how my content is used in AI systems.
I never gave consent to use my content as training data for AI. As this is not personal data, and most of my blog posts are sharing know-how, I have no problem with that.
But, the shared know-how is never sourced with my name or Fresh van Root, and that is a problem long-term. So, at least for now, AI chat systems are a one-way street, not sending any traffic to your site.