My Google Gemini AI Research Rabbit Hole begins
I asked Google Gemini '2.5 Flash' on July 11, 2025, "What is BoxScoreDiceSports.com?"
Honestly it wasn't that long ago that these Large Language Models (LLMs) didn't do so well with contemporary information. But the AI landscape is changing rapidly. Here is what I got...
What is Boxscoredicesports.com
BoxScoreDiceSports.com appears to be the website associated with Box Score Dice Sports, a creator of dice-based sports simulation games.
Based on the information available, here's what we can gather about it:
- Creator: The games are created by an individual named Jason Staben, based in South Australia, Australia.
- Purpose: The website and associated profiles (like on Pinterest and Etsy) are where Jason shares and sells the sports games he has developed.
- Game Style: These are tabletop, pen-and-paper games that utilize dice rolls to simulate various sports. The focus is on producing realistic statistics and results.
- Available Games: Examples of games mentioned include:
- Dice Cricket (including Test Matches, Limited Overs, and T20)
- Aussie Rules Dice Footy
- Inning Baseball (versions 1 & 2)
- Series Baseball
- Format: The games are typically sold as digital downloads (ZIP files) and include instructions, game charts, scorecards, and often bonus Excel spreadsheets for generating skill ratings or random run rate cards.
- Availability: While the website Boxscoredicesports.com is listed, the actual sales and detailed descriptions of the games appear to be primarily through an Etsy shop named "BoxScoreDiceSports".
In essence, Boxscoredicesports.com is the online presence for a passionate creator of sports simulation dice games, offering a unique way to enjoy and analyze sports through a tabletop format.
Okay, not too shabby. couldn't have said it better myself, thanks Gemini.
I then prompted it to "Create a image based on Boxscoredicesports style and theme."
This image does not represent what I create. But hey, looks nice.
Overall, it was a pretty easy prompt for a LLM to answer in 2025, but at the sametime, I am still impressed with what they are able to do these days. For example, I 'Vibe Coded' this Browser based Word Processor (it took about 30 mins of prompting). I don't recommend using it for anything serious, but it certainly does work.
This got me thinking, Maybe I could feed prompts about some of my games and see what I get. It might be interesting... And, if it works, I will share them on this site in the future.
Another good use might be getting skill ratings for games. I know that some people already do this and good on them! I am far from an expert, but I have learnt pretty quickly that the results are often only as good as the prompts fed into the LLM.
My attempt to get 2024 Buffalo Bills Skill Ratings.
I chose the 2024 Buffalo Bills roster. Why? It was literally the first team I clicked on.
My prompt: "based on the 2024 Buffalo Bills NFL roster and stats found at https://www.pro-football-reference.com/teams/buf/2024.htm please rank each player out of 1 to 10 as well as allocated them the main position (i.e. QB, WR, RB) and maybe a backup position they play. Put the results in a HTML formated table with column headings of Name, Age, Pos, Games, Skill."
Gemini's initial response was basically, no I can't do that. It wrote:
"I was able to access the 2024 Buffalo Bills NFL roster information from the provided link. However, the data on the webpage is structured in a way that makes it challenging to automatically extract a comprehensive list of all players with all the requested statistics and potential backup positions in a single, parsable format..."
It went on, but you get the idea.
It did suggest I try letting it do deep research, just by pressing the button to being it! So, naturally I did (all of this is the free version by the way).
So I proceeded and after about 10 minutes of researching it came back with a comprehensive analysis of the Bills 2024 season. It included a ranking of each player out of 10 as I had requested.
What I got blew me away 🤯🤯🤯
Click each link to view below
- 2024 Season Interactive Analysis
- Infographic
- Full Report (PDF)
- Audio analysis between two AI commentators
- Season Quiz - Opens a new TAB
How do humans compete with AI?
I must admit I do not know enough (or anything really) about the Bills to know if this is good analysis. I will leave that up to a Bills fan. Regardless, this report has been put together with little effort by me, just a desire to rank the 2024 players out of 10. The depth of this report is incredible. I especially love the audio commentary.
Conclusions... for now
If you are reading this then I am potentially still adding thoughts to this article, but I will wrap up my thoughts for now. Bascially, AI is an incredible tool that can aid us in our gaming experience. I think that if you enjoy the sort of games I offer here, then things are not always about automation. There is something about the process and the journey. We could get AI to make a game engine that spits out seasons of games quickly (and honestly I think I will try it), but in doing that you can miss the reward of individual games and individual moments in those games that stick in your mind for years to come. At the same time, there are some jobs, such as rating players that can be tiresome, if AI can knock out a roster for us qucikly based on deep analysis then all the better!