4 Weeks, 3 Games, and 1 AI-coach: What AI Taught Me About Gaming

July 30, 2026

After ~4 weeks of gaming, I am happy to finally conclude this experiment.

Before this month, I was (and still am) a casual gamer. I enjoyed the challenge of some games and the stimulation a well-thought-out design brought about. To me, complex problems, strategies, and obstacles were fun when under the premise of low-risk environments, as–after all–they were just games.

Games provided an avenue for cognitive exercises whilst eliminating the risks and dangers of doing them in real life. “Want to role-play as a General in times of War?” Check out Command and Conquer. “Interested in building your own world from scratch?” Here is Minecraft!

In the 21st century, games are abundant and tailored to a wide variety of tastes for anyone and everyone to play. And every year, innovations come out, like the popular VR (Virtual Reality) and AR (Augmented Reality) games.

But just as the gaming industry has progressed and continued to progress, it did make me wonder: where can AI come in? While yes, AI can help streamline the development of games in terms of software development and even in the creative aspect, it got me thinking where else AI can come into play. That thinking led me to my experiment: Playing games and using Generative AI as my coach (specifically Gemini).

The Premise

At the beginning of my experiment, I wanted to figure out the following: (1) Does an AI Coach increase my chances of winning the game or successfully completing the tasks?, (2) Would an AI Coach teach me how to more effectively "game"?, and (3) Does the added presence of an AI coach add to my enjoyment of the game?. Fundamentally, I wanted to assess how well AI could be as a gaming coach and how far I can go with it. Additionally, I also wanted to explore whether AI is better or worse at specific kinds of games, as well as whether AI would help in flattening the learning curve for new games.

Results

From the three games I played, the AI-provided strategies and suggestions definitely increased my chances of winning substantially. AI’s knowledge of Meta strategies and robust statistical computations made beating “solved” games (i.e., games whose outcomes can be perfectly predicted from any given position, assuming fixed conditions and an ideal number of moves) way simpler. Whilst the games I had played were not as fixed as, say, a puzzle or Sudoku, understanding the mathematics behind the game structure would give you an advantage (and AI is great at that).

Essentially, in playing with AI support, I was inclined to make the most ideal moves every turn, which led to an increased success rate. Having a buffed computational brain (with the AI) allowed me to identify more beneficial choices and moves that I would have missed if I had been playing the game as usual.

However, later on, it seemed like each of my gaming sessions followed a similar pattern when playing with the AI. In the initial rounds, I would be more actively prompting Gemini on what the best strategies were, what tactics I should keep in mind, and it went even as far as asking what I should do at every turn. But at a certain point in every session, my prompting would die down, and I would be more focused on playing the game itself. Per game, I would end up prompting AI only 1 to 3 times, with the majority of the prompting being done pre-rounds.

And after playing 3 different games, and recording a total of 6 hours of game time, I was able to gather several observations on the major differences between playing with and without an AI coach.

Takeaways

AI played the role of Coach so well that I began to assume the role of an ideal player–executor of its will, if you will. But later on, as the AI began to run out of new strategies or insights to give me, that's when my sense of autonomy started to kick back in.

Kind of like learning how to drive for the first time, normally you start out with hands tightly gripping the wheel, eyes laser-focused, and listening to every directive your parent or driving instructor tells you. But as you get the hang of maneuvering the vehicle, getting your bearings and spatial awareness on, you begin to ease up–asking your instructor fewer questions and just doing your thing.

That is in essence what the experience of having an AI Gaming Coach felt like. While AI is insanely intelligent, when it comes to these games, it eventually hits a wall where it can’t offer you any new knowledge. Would the experience be different had I played different genres of games? Most probably, yes. But in the outcome of this experiment, AI is more than adequate, and maybe even overkill, in completing “solved” games.

Overall, AI is a great coach for those trying to improve in tactical strategies and learn the current Meta strategies for games similar to Catan, Balatro, and 9 Kings. If you want to understand how to analyze what the ideal strategies are for more “solved” games, Gen AI could definitely be an effective teacher.

However, for those like me who enjoy the journey of figuring out the game and care less about high scores and choosing the ‘ideal’ moves everytime, I definitely would tell you to skip the AI coach and just play the game. No one likes micromanagers (unless you don’t know what you're doing) so just solve the game yourself and take all the credit!

FINAL VERDICT

Learning new ways to beat the game: 4/5 🏆🏆🏆🏆
"It-felt-like-cheating" Rating: 3/5 👀👀👀

Link to Gemini Chat:
https://share.gemini.google/Y8zAPtXoMi3m

brb gonna play another game,

Ella