Unveiling the hurdles faced while crafting initial requests to OpenAI's GPT-3 for my medieval game
Venturing into the world of OpenAI's GPT-3.5 for generating a dynamic narrative for my game was akin to stepping into a medieval fantasy, filled with promise yet fraught with challenges. My initial foray was filled with the enthusiasm of a bard entering a bustling tavern, ready to regale the crowd with tales of heroics, courtesy of GPT-3.5. My basic prompt to the API was simple: "As a bard of a medieval tavern, tell a story about a hero, provide for me 3 choices." The idea was to continue the narrative based on the choices provided, a cycle of request and response that would spin the yarn of an epic tale.
The maiden voyage was smooth, but soon, the dragons of reality began to rear their heads. Here’s a recount of the challenges that ensued:
The antidote to the first two dilemmas was to craft a skeleton for the narrative. A core structure of locations and characters, coupled with additional information sent with each request, aimed to guide the narrative like a compass. However, this solution unfurled a new challenge, a hidden trap in the quest that awaited my next move. The tale of tackling this newfound challenge is a story for another day, a page in the chronicle that I am eager to share in the subsequent chapter.
As I delve deeper into the realms of coding, OpenAI, and narrative generation, each challenge unfurls a lesson, a step closer to the holy grail of creating a captivating game of choice set in the medieval era. Stay tuned for the next chapter where I’ll unravel more of this intriguing expedition from the realm of Product Management to the hands-on world of Coding and Game Development.