Having spent considerable time exploring machine learning and developing business applications using such techniques during my career, the broadening application of artificial intelligence to search engines and expert systems is fascinating. General-use applications such as ChatGPT, Gemini, and CoPilot have taken the techniques of natural large language processing (nllp) models and offered them to the masses.
Intrigued by recent AI experiments by two wargaming bloggers,
I buckled in, fired up the search engine, and posed the following question to perplexity.ai:
I asked, What is the focus of the Palouse Wargaming Journal?
After only a few seconds wait, the algorithm returned the following summary:Based on the search results, the focus of the Palouse Wargaming Journal blog appears to be:- Wargaming and miniature wargaming, covering a variety of historical periods and conflicts such as the Italian Wars, Napoleonic Wars, American Civil War, etc.
- Providing battle reports and game recaps of wargaming sessions played by the blog author and his gaming group. The blog frequently features posts about specific wargames played, such as the "Battle of Sanguetta 1497" series.
- Discussing the author's painting and modeling projects related to his wargaming hobby, including updates on figures and terrain he is working on.
- Sharing the author's thoughts and reflections on wargaming, game design, and the broader wargaming community. For example, the post "Thoughts on Re-Fighting History: War in the East" discusses the evolution of wargaming mechanisms and content.
- Engaging with the wargaming community, as evidenced by the blog's participation in book giveaways and discussions.
In summary, the Palouse Wargaming Journal appears to be a blog focused on the author's personal wargaming activities, battle reports, painting projects, and commentary on the broader wargaming hobby.
The results are not too bad for a general assessment of the content and focus of this blog. In fact, the results are quite good.
Notice that the emphasis on Italian Wars content seems to score higher than older content especially when the recent series of battle reports on Sanguetta are singled out. Since the battle reports on Sanguetta have only been published in March, search currency suggests that these nllp models are being retrained frequently. The Thoughts on Re-Fighting History scored highly too. Could this result be driven by reader commentary and discussion as (5) suggests? Are these models devouring content as well as commentary?
Is Google actively vacuuming up Blogger content (without permission) to train its large language models? Would this activity lead to an increase in pageview traffic or remain unseen in Blogger statistics?
This experiment raises a number of questions, hopefully, without making me out as a conspiracy theorist. I may re-run this experiment later and see if results change as content evolves.
Have you tested these waters with your own blog? What are your results?