As generative AI becomes more prevalent, the market for prompt trading is rapidly expanding. However, current platforms like PromptBase still primarily dictate pricing from the seller's perspective, lacking an objective standard for price measurement. Facing this challenge, the Multimedia and Intelligent Security team at Fudan University has proposed an innovative model for prompt trading, aiming to better adapt to future buyer-oriented markets.
This new trading model consists of two main phases: prompt category selection and pricing strategy formulation. In the first phase, the platform employs a multi-armed bandit algorithm based on greedy search to select prompt categories for sale based on quality assessment. The second phase utilizes a cascaded Stackelberg game approach, treating buyers, platforms, and sellers as primary leaders, secondary leaders, and followers respectively, with a focus on buyer interests.
The core of this model is to comprehensively consider the relevance and quality of prompts to generated content, allowing all parties to formulate optimal strategies after balancing costs and revenues. By setting reasonable price ranges and requirements for prompt richness, this model effectively balances the interests of all three parties, potentially leading to a win-win situation.
Researchers Li Meiling and Ren Hongrun detailed this model in a recent paper published on arXiv. They believe that this trading model not only better regulates the prompt market but could also reduce the costs for content creators and enhance their efficiency.
With the increase in the number of prompt goods and the reduction in transaction costs, this model is expected to reshape the ecosystem of AI content creation. However, the research team also points out that the design of profit functions for all parties and the assessment of prompt quality are still key factors affecting the final pricing. In the future, they plan to extend this achievement to a broader range of prompt pricing scenarios.
This research provides new insights for solving the prompt pricing conundrum and is expected to play a significant role in future AI content creation and trading.
Address: https://arxiv.org/pdf/2405.15154
