Bill Gross’s ProRata Aims to Pay AI Creators Fairly

Bill Gross, a veteran Silicon Valley founder and investor, is attempting to restructure the economic relationship between generative AI companies and content creators through his startup, ProRata. The platform aims to establish a market where publishers and creators are compensated based on how their work informs AI-generated outputs.

The initiative follows a period of increasing tension between AI developers and media organizations, highlighted by lawsuits such as the one filed by The New York Times against OpenAI. Gross argues that while litigation is one path toward resolution, a sustainable business model is a more effective solution.

ProRata has signed 1,500 publications over the last two years. The company’s proposed model involves sharing 50% of revenues with creators, utilizing a patented attribution method to determine exactly which sources contributed to a specific AI-generated answer.

Gross acknowledges that no AI operators are currently paying through ProRata, describing the effort as a long-term strategy. He believes AI companies will eventually be forced to adopt such models once they lose existing lawsuits, achieve profitability, or find that the quality of their answers declines due to a lack of current information.

A central tenet of Gross’s approach is the distinction between the use of content for training and its use in final outputs. He suggests that while crawling content to train a model may be acceptable, the actual use of that content to generate a response should trigger payment.

Gross advocates for a dual-payment system where publishers receive compensation for both the input phase—crawling and training—and the output phase. He notes that some publishers, such as The Atlantic, already have arrangements with OpenAI that cover these aspects.

To provide immediate value to publishers while waiting for AI operators to adopt the ProRata model, Gross launched a spinoff called Gist. This tool helps publishers monetize their content more effectively by leveraging the data gathered during ProRata’s indexing process.

Gist provides AI-powered site-specific search designed to outperform traditional keyword search by understanding the context and knowledge graph of articles. The tool also suggests related questions to users, which Gross reports results in click-through rates between 3% and 5%.

Gross believes this model is particularly beneficial for smaller publishers and vloggers. He argues that in an AI-driven environment, the quality and uniqueness of content are more important than brand size, allowing the long tail of creators to receive proper compensation.

The development of ProRata and its products has relied heavily on the very technology Gross is seeking to regulate. He notes that a generative engine optimization product was built by 13 people in four months, a task he estimates would have required 30 people over four years using traditional methods.

Gross also detailed using ChatGPT to resolve a dispute with the patent office. After the office rejected 13 of 15 claims in a patent application, Gross used the AI to help synthesize a more convincing explanation of why the claims were valid, which subsequently led to the approval of all claims.

Industry Outlook and AI Profitability

Regarding the broader AI market, Gross believes that while current valuations are high relative to profits, they are not necessarily indicative of a bubble because revenue growth is substantial. He describes the current state of Large Language Models as possessing a useful, though statistical, “fake understanding” of information.

Gross offered varying assessments of the major AI players:

• Anthropic: Gross suggests Anthropic may reach profitability sooner than its peers by focusing on enterprise clients and throttling usage to manage costs.
• OpenAI: He believes OpenAI is shifting toward a model that includes advertising to monetize its large consumer base.
• Meta: Gross characterized Mark Zuckerberg’s AI and metaverse investments as costly ventures that may not yield the same success as Instagram, noting that Meta is not currently in a leading position.
• xAI: While he views Elon Musk’s general direction as correct, Gross claims Musk frequently exaggerates development timelines by about 10 years as a fundraising tactic.

Looking toward the next decade, Gross predicts that AI will fundamentally upend society. He expressed concern that the benefits of this shift may not be distributed evenly, potentially increasing the wealth gap if the value does not flow down to the creators providing the underlying data.

The right way to solve this isn’t with lawsuits. The right way to solve This represents with a fair business model, because a fair business model would enable everybody to participate properly and you wouldn’t need to sue anybody.

Bill Gross