# Should ChatGPT and Bard Share Revenue with Their Data Providers? A New Business Model for the AI Era

**Authors:** Dong Zhang  
**arXiv:** [2305.02555](https://arxiv.org/abs/2305.02555)  
**Date:** 2023-05-04  
**Summary page:** https://www.summarizepaper.com/en/paper/2305.02555v1/

## Key points

- AI tools such as ChatGPT and Bard require more and better quality data to continuously improve
- Sharing revenue between AI tools and their data providers could transform the current hostile zero-sum game relationship into a collaborative and mutually beneficial one that drives forward AI technology and builds a healthy ecosystem
- Current revenue-sharing business models do not work for AI tools in the forthcoming AI era since new metrics such as prompts and cost per prompt for generative AI tools will replace traditional website-based traffic and action metrics
- A completely new revenue-sharing business model needs to establish a prompt-based scoring system to measure data engagement of each data provider
- The proposed scoring system would encourage more data owners to participate in the revenue-sharing program, creating an environment where all parties benefit
- This model must be almost independent of AI tools and easily explained to data providers, applying not only to large language models but also other types of multimodal AI tools like text-to-image generators or those used in healthcare.
- Future work should focus on refining this model further so that it can be applied across different sectors beyond language models.

## Summary

As AI tools such as ChatGPT and Bard become increasingly popular, the question arises of whether they should share revenue with their data providers in addition to traditional stakeholders and shareholders. The answer is yes, as large AI tools always require more and better quality data to continuously improve. However, current copyright laws limit their access to various types of data. Sharing revenue between AI tools and their data providers could transform the current hostile zero-sum game relationship into a collaborative and mutually beneficial one that drives forward AI technology and builds a healthy ecosystem. Current revenue-sharing business models do not work for AI tools in the forthcoming AI era since the most widely used metrics for website-based traffic and action will be replaced by new metrics such as prompts and cost per prompt for generative AI tools. A completely new revenue-sharing business model needs to establish a prompt-based scoring system to measure data engagement of each data provider. This paper systematically discusses how to build such a scoring system based on classification and content similarity models, outlining the requirements for third parties or tool developers to build it. The proposed scoring system would encourage more data owners to participate in the revenue-sharing program, creating an environment where all parties benefit. However, this model must be almost independent of AI tools and easily explained to data providers. For example, it could apply not only to large language models but also other types of multimodal AI tools like text-to-image generators or those used in healthcare. In conclusion, sharing revenue with data providers using an effective scoring system would facilitate collaboration among all stakeholders involved in developing exceptional AI tools that reap considerable profits while building a healthy ecosystem. Future work should focus on refining this model further so that it can be applied across different sectors beyond language models.