As artificial intelligence (AI) becomes more powerful, one pressing issue is how the data used to train AI models is sourced, shared, and managed. At India Blockchain Week 2024, Avery Ching, co-founder and CTO of Aptos, highlighted a novel way blockchain could address this challenge: giving creators clear, verifiable consent over how their content is used to train AI.
In a world where large-scale AI models are often trained on vast amounts of publicly available data—sometimes without the content creator’s consent—Ching sees blockchain as the ideal solution for building transparency, control, and fairness into this process.
Let’s break down why AI training consent is the “perfect use case” for blockchain, according to Ching, and what this could mean for the future of both AI and decentralized technologies.
AI Training Data: The Missing Link in Consent
The rise of AI, particularly large language models (LLMs), has sparked a conversation around data privacy and ownership. AI models like ChatGPT, Google’s Bard, and others are trained on vast datasets sourced from a wide array of content: books, websites, social media posts, and more. But what happens when that content belongs to creators who haven’t agreed to let their work be used for AI training?
Avery Ching sees this dilemma as a “perfect use case” for blockchain technology. Blockchain, with its immutable and transparent nature, can provide a solution to the issue of consent in AI training.
“As we continue to see AI develop, there’s going to be a growing need to ask: do people want their content to be used for AI training?” Ching said during his talk at the event. “Blockchain offers a way to give content creators control over whether or not their material is used to train these models.”
How Blockchain Could Revolutionize AI Training
So, how exactly could blockchain solve the consent issue in AI training? According to Ching, blockchain’s immutable ledger offers the perfect solution. When content creators upload their work to a blockchain, they could attach consent conditions for how their content is used—whether it’s for training AI, sharing data, or any other purpose.
Blockchain would essentially serve as a decentralized permissioning system, where every creator has full control over how their work is accessed. This system would be built around smart contracts that automatically enforce the terms of consent. For instance, creators could specify that their content can only be used for non-commercial AI training or that it can only be used with certain crediting or remuneration.
By ensuring that AI models are only trained on content for which creators have explicitly given their consent, blockchain could play a pivotal role in addressing ethical concerns around data usage in AI development.
“It’s about providing a center of control for creators, where they can say ‘yes, I want my content to be used for this purpose,’ or ‘no, I don’t.’ And this control can be verified and enforced without relying on a central authority,” Ching explained.
Challenges of Scaling AI Consent on Social Media
While blockchain offers a powerful solution, scaling this consent system presents some technical challenges. Ching, who has a background in tech at Meta (formerly Facebook), pointed out the sheer volume of content generated daily on platforms like Facebook, Instagram, and Twitter.
Social media platforms generate billions of pieces of content every day, and applying blockchain-based consent controls to each of those data points would require high transaction throughput and could incur significant costs. For each post, image, video, or comment, the blockchain would need to record its consent status, potentially making this a resource-heavy task.
“There’s a scalability challenge,” Ching acknowledged. “If you’re going to add controls around each piece of data, it’s going to involve a lot of transactions, and that can be expensive. This is especially true when you consider platforms with billions of users generating vast amounts of content every day.”
Aptos: A Blockchain Solution for AI Consent
Despite these challenges, Aptos is actively working on a blockchain solution that could support AI consent use cases at scale. As a next-generation blockchain built for scalability, speed, and security, Aptos is positioning itself as a potential platform for managing AI training consent.
Aptos leverages a unique consensus mechanism and technology stack that could handle the high transaction throughput required to manage consent for large-scale content distribution. The Aptos blockchain is designed to support complex applications and decentralized ecosystems—exactly what’s needed for a system that manages AI training permissions.
As AI continues to grow and become more sophisticated, the need for decentralized control over the content used to train these models will only increase. Ching sees blockchain as the natural backend for managing this, offering a transparent and automated way to track content usage, manage permissions, and ensure that creators are fairly compensated or credited for their contributions.
The Bigger Picture: Ethical AI and Decentralization
The potential to combine blockchain and AI training consent aligns with broader efforts to make AI more ethical and transparent. As the AI industry faces increasing scrutiny over data privacy, ownership, and biases, technologies like blockchain could play a crucial role in ensuring that AI development is done with fairness and accountability in mind.
For Ching and Aptos, this is more than just a technical challenge—it’s about creating a new ecosystem where creators are empowered and have the final say over how their content is used. Blockchain, with its transparency and immutability, could offer the kind of decentralized governance that’s necessary to reshape how we think about AI ethics.
Looking Ahead: Blockchain’s Role in AI’s Future
The intersection of blockchain and AI is a growing field, and Aptos is one of the many blockchain projects exploring ways to improve AI training practices. While we may be some time away from a fully decentralized system for managing AI consent, Ching believes this is an important step forward in ensuring that the rise of AI doesn’t come at the expense of creators’ rights.
“We’re not just talking about a new technology here,” said Ching. “We’re talking about creating a new framework for how AI and blockchain can work together to empower creators, protect their rights, and provide a more equitable way to use content for training.”
As AI technology evolves, the need for blockchain-based consent mechanisms will only grow stronger. It’s clear that in the near future, blockchain could be the key to managing ethical AI and ensuring that creators maintain control over their content in a decentralized, transparent way.
With Aptos and other blockchain platforms pushing for innovation in this space, the idea of ethical AI driven by user consent could soon become a reality—one that’s built on the principles of decentralization, transparency, and fairness.
