A Hands-On Method for Developing Creative Content and AI Learning
Artificial intelligence is accelerating progress in significant ways, transforming everything from our daily routines to the forefront of scientific discovery and creativity. As AI advancements gather pace, it becomes essential to promote a balanced approach to the use of creative content in training AI models. This challenge extends beyond legal matters, touching on the future of AI innovation and human creativity.
Every new technology for the creation or transmission of knowledge and art, from the printing press to the internet and cable TV, has raised questions about how to create and share value. In the context of AI, developers have numerous opportunities to support the creative industries and cultivate a thriving AI ecosystem that benefits all stakeholders. What approaches are most effective regarding the outputs of AI models, the training methods employed, and the novel ways AI can generate shared value?
Assessing AI Outputs
Whether words are generated with a pen, typewriter, or AI, or artwork is created with a paintbrush, computer graphics, or AI, determining if a new work infringes on the copyright of an earlier creation can be complex. This evaluation depends on several factors, including the similarity between the new work and the original, the nature of both works, and market competition. Tools such as output filters can help minimize substantially similar outputs, even as models learn to make more nuanced judgments.
Moreover, incorporating provenance information, such as watermarks or metadata, can mitigate the risk of misrepresentation regarding the creator of specific material. Initiatives in the industry are already underway to help consumers make informed evaluations about the content they encounter.
Training AI Models Responsibly
While training foundational AI models on content accessible from the open web is considered a transformational fair use under U.S. copyright law, and several other countries have similar provisions, adhering to good practices can enhance acceptance of new AI applications using existing content.
Acquiring content responsibly and lawfully is crucial, including providing web sites with the option to opt-out of permitting their content for AI training. Established industry standards that govern web crawling, built on the widely recognized robots.txt protocols, offer a practical approach to achieve this. Many web publishers are now utilizing AI-specific protocols to further refine their control over content accessibility. AI developers should remain adaptable to evolving standards as the ecosystem matures and take appropriate steps to avoid improperly training general-purpose AI models in ways that circumvent these standards.
To prevent misuse of individuals’ voices and likenesses, legislative frameworks should enhance existing “notice-and-removal” systems for copyright, incorporating essential safeguards. New tools can empower creators to harness AI’s creativity while maintaining control over their personal attributes.
Sharing Value, Expanding Opportunity
AI holds the potential to benefit everyone. Collaboration between AI developers and content publishers can broaden the market and create new income streams for creative industries. Developers are exploring ways to share the value generated by outputs by directing related traffic to content providers. Together, the ecosystem is identifying new methods to tap into value from emerging AI applications.
For instance, commercial partnerships may arise when AI services base their responses on factual information from various websites. Developers and content publishers are also negotiating new agreements for utilizing specialized or non-public data for training purposes, with developers increasingly discerning the relevance of individual content for different applications.
AI developers are committed to collaborating with media and creative industries to create innovative generative AI tools that enhance these fields. Tools like Pinpoint, designed for journalists, aid reporters in sifting through text, audio, image, and video files to uncover data patterns, identify fresh angles, or locate quotes within media files.
AI represents a shared opportunity that has the potential to expand the horizons of science, commerce, and creativity. We are dedicated to working with all stakeholders within the ecosystem to establish a collaborative framework that promotes both creators’ rights and innovation.
