I tested whether ChatGPT could beat Slay the Spire 2…. I was very surprised.

Exploring AI’s Capabilities: Can ChatGPT Conquer Slay the Spire 2? A Surprising Experiment

In the rapidly evolving intersection of artificial intelligence and gaming, enthusiasts and developers alike are eager to explore how AI models like ChatGPT perform in complex strategic environments. Recently, I undertook an intriguing experiment to test whether ChatGPT could successfully navigate the challenging deck-building roguelike, Slay the Spire 2. The results were unexpectedly enlightening.

For this experiment, I tasked ChatGPT with handling all key decision-making processes during a full run of the game. This included selecting cards, managing shop purchases, and determining strategic priorities. The AI’s choices ranged from seemingly erratic to surprisingly sound, with some decisions appearing to lack logical basis, while others demonstrated a noteworthy understanding of the game’s mechanics.

Admittedly, I entered this challenge with modest expectations. However, the outcome surprised me in many ways. The AI managed to make a series of moves that, at times, aligned well with optimal strategies, despite occasional questionable choices. It was fascinating to observe how an AI language model interpreted game states and formulated decisions in a context where nuanced judgment often makes the difference between victory and defeat.

Would you have made the same decisions if you were in the AI’s place? The experiment raises interesting questions about the current and future capabilities of AI in strategic gaming contexts. While AI still struggles with consistent optimal play, experiments like this highlight its potential to understand and adapt in complex decision-making environments.

If you’re curious to see how ChatGPT performed firsthand, you can watch the full run in this video: Can ChatGPT Beat Slay the Spire 2. It’s an engaging look at the intersection of artificial intelligence and gaming strategy, showcasing both its limitations and surprising strengths.

As AI continues to develop, experiments like this not only challenge our assumptions but also open up exciting possibilities for integrating AI into gaming, game design, and strategic decision-making in the future.