Understanding MARTI: A New Metadata Framework for AI 19 Nov 2024, 4:08 pm

At its core, MARTI is a bridge. It harmonizes with existing metadata standards like the Content Authenticity Initiative, Anthropic’s Responsible Scaling Policy, and the W3C’s PROV. It anticipates the needs of future standards, laws and practices, such as those proposed by the Coalition for Networked Information (CNI), The EU Artificial Intelligence Act, and Making Data FAIR.—Carrie Bickner

As I study Carrie Bickner’s initial posts on the MARTI Framework she’s developing to manage AI metadata across various disciplines, a familiar feeling steals over me.

It’s similar to how I felt during the early days of The Web Standards Project (WaSP), when a handful of us took on the quarreling browser makers in what seemed a Quixotic attempt to bring consistency, predictability, usability, and accessibility to an already Balkanized web.

Fortunately, at that time, we had two aces up our sleeves: 1., the standards already existed, thanks to the W3C, and 2., the EU and Clinton Administration were suing Microsoft, which meant that the tech press was interested in hearing what we had to say—even if evangelizing web standards had little to do with accusations that Microsoft was abusing its monopoly power.

Once more with feeling: standards from the community

Years after The WaSP declared victory, and browser stagnation had begun to set in, I felt that same thrill vicariously when Eric Meyer, Tantek Çelik, and Matt Mullenweg invented XFN (XHTML Friends Network), inverting the standards creation pyramid so that great ideas were empowered to bubble up from small groups to the wider community, Open Source style, rather than always coming from the top (W3C) down.

I’ve no doubt that microformats were the spark that lit the HTML5 fuse, and we all remember how Steve Jobs used the new markup language to power the first iPhone, initiating the mobile era we now live in.

More about microformats history is available, and you can read Jeremy Keith’s HTML5 For Web Designers online for free—or buy the 2nd Edition, coauthored with Rachel Andrew, directly from Jeremy.

And now I feel those same stirrings, that same excitement about possibilities, as I study Carrie’s first posts about MARTI, an emerging object-oriented metadata framework that can be used to articulate rights-permissions, preservation metadata, provenance, relationships between objects, levels of AI involvement, and contextual information such as usage history and ethical considerations. 

Here’s why I’m excited (and you may be, too).

What do you wanna do tonight, MARTI?

For better or worse, our ideas create our reality. For better or worse, we have atomic power, the web, and social media. There’s no putting these genies back into their bottles. And there’s certainly no shutting down AI, however you may feel about it. Nor need we, as long as we have smart guardrails in place. 

I believe that MARTI—particularly as it promotes responsibility, transparency, and integrity in documenting AI’s role in content creation and curation—has the potential to be one of those guardrails.

Drafted by a career digital librarian, this provisional  metadata framework for human/generative AI output won’t stop bad actors from scraping content without permission. But if it is extended by our community and embraced by the companies and organizations building AI businesses, MARTI has the potential to bring rigor, logic, and connectedness to the field. In Carrie’s words:

The emergence of generative AI marks a transformative moment in human creativity, problem-solving, and knowledge-sharing. MARTI (Metadata for AI Responsibility, Transparency, and Integrity) is a provisional metadata framework designed to navigate this new landscape, offering a standardized yet adaptable approach to understanding, describing, and guiding the outputs of human-AI collaboration—and even those generated autonomously by AI.

At the heart of MARTI lies a robust object model—a modular structure that organizes metadata into reusable, interoperable components. This model ensures transparency, traceability, and ethical integrity, making it the cornerstone of the MARTI framework.

MARTI is not just an architecture for describing AI output, but it offers a way of structuring policy and a possible foundation for a new literacy. This is not about teaching every individual to code or engineer prompts. It’s about empowering humanity to collectively understand, describe, and guide everything we make with AI, ensuring accountability, transparency, and ethical integrity at every step.

MARTI is a framework for creating structured, standardized documentation that is attached to or embedded in AI-generated content. This documentation, or metadata, can be created by people collaborating with AI tools to produce content. Additionally, AI processes themselves can generate and embed metadata into their outputs, ensuring transparency, traceability, and accountability at every stage of content creation.

MARTI also offers a variety of potentially transformative business applications.

Disclaimer: the author is a friend of mine. But then again, so is every other thought leader mentioned in this article (with the exception of the late Steve Jobs, although our lives did touch when he fired me from a project—but that’s another story).

For more MARTI magic, check these posts:

• Meet Marti
• Object Model Overview
• Domain-Specific Attributes in MARTI
• MARTI: The Object of Our Affections
• New! Safeguarding Intellectual Property: How Blogging Platforms Can Use MARTI to Protect the Creative Content Ecosystem
• Also new! Help Us Establish a Home for MARTI
• Haikus, Sonnets, and Leaves: The Morphology of Metadata for Intellectual Units
• This Is Your Brain on ChatGPT

And if you’ve a mind to do so, please pitch in!

The post Understanding MARTI: A New Metadata Framework for AI appeared first on Zeldman on Web and Interaction Design.