A decentralised AI is a technology at the intersection of emergent tech including blockchain, AI and edge computing (IoT). What are the potentials of decentralised AI, its benefits, and how it can shape the future of technology? These are some of the questions that Dr. Paolo Tasca answers in the recent Dinis Guarda YouTube Podcast.
Currently, the larger share of the AI market is being controlled by big tech giants including Microsoft, Google, and IBM, that offer cloud-based solutions and APIs. This central model not only limits transparency and interoperability, but it also leads to unfair pricing and ultimately the monopolisation of the AI market.
Now imagine a world where every participant in a network has equal power and a true ownership of their data. This model is what we call the decentralised artificial intelligence (DAI). Decentralised AI has the potential to revolutionise industries, and enhance data privacy, while offering a democratic access to AI-driven solutions.
"We are already seeing a pervasive growth of AI, a solution that is providing to almost every aspect of our daily lives. What I see is the emergence of a new dimension: a decentralised AI, which has the possibility bringing AI components to the distributed networks", Dr. Tasca told Dinis in the interview.
Let us explore more about Decentralised AI:
The convergence of blockchain and AI: Decentralised AI
Blockchain technology is renowned for its secure, immutable, and transparent ledger capabilities. AI, on the other hand, empowers machines with learning and decision-making abilities. When these two transformative technologies converge, they create a powerful synergy that holds the potential to disrupt traditional centralised AI systems.
Decentralised AI, at its core, involves the utilisation of blockchain networks to facilitate AI model training, deployment, and data sharing. Here's how it works:
Data Privacy: In centralised AI systems, user data often resides in the hands of a few powerful entities, raising concerns about data privacy and misuse. Decentralised AI allows individuals to retain control over their data by enabling them to securely share it on blockchain networks without revealing sensitive information. This ensures that users have ownership and autonomy over their data.
Enhanced Security: Blockchain's cryptographic features make it exceptionally secure. Decentralised AI leverages this security to safeguard AI models and data. It reduces the risk of data breaches and unauthorized access, a critical concern in the digital age.
Fairness and Transparency: Traditional AI models can suffer from biases, leading to unfair outcomes. Decentralised AI, built on transparent blockchain networks, can help address these biases by providing a clear audit trail of data sources and model decisions. This transparency fosters fairness and accountability.
Incentivised Collaboration: Blockchain networks often use tokens to incentivise participants. In decentralised AI, data providers, model developers, and users can be rewarded with tokens for contributing to the ecosystem. This encourages collaboration and innovation.
Use cases of Decentralised AI
Healthcare: Patients can securely share their medical data with AI models for diagnosis and treatment recommendations without compromising their privacy.
Companies like Solve.Care were working on blockchain-based healthcare platforms that aimed to give patients control over their medical records and enable secure data sharing with healthcare providers.
Finance: Decentralised AI can enhance fraud detection, automate trading strategies, and provide personalised financial advice while keeping sensitive financial data secure.
Supply Chain: Tracking and verifying the authenticity of products throughout the supply chain becomes more transparent and efficient with decentralised AI.
IBM's Food Trust platform was using blockchain and AI to enhance transparency in the food supply chain, allowing consumers to trace the origin of food products.
Energy: Decentralised AI can optimise energy consumption, enable smart grids, and facilitate efficient energy trading on blockchain networks.
The Energy Web Foundation was working on blockchain-based solutions for decentralised energy grids, allowing for more efficient energy distribution and management.
Content Creation: AI-driven content creation platforms can reward content creators and consumers with tokens, ensuring fair compensation and quality content.
Challenges and considerations with Decentralised AI
Despite its immense potential, decentralised AI faces challenges such as scalability, energy consumption, and regulatory compliance.
"When we put AI on blockchain, we amplify the risks of AI for the world", said Dr. Tasca.
One of the primary concerns is the need for robust governance and consensus mechanisms within decentralised AI networks to ensure ethical use, accountability, and prevent malicious activities.
Additionally, managing the scale and complexity of decentralised AI systems can be daunting, requiring efficient algorithms and infrastructure. Interoperability between various decentralised AI protocols and systems is another hurdle, as seamless communication and data exchange are essential for their effectiveness.
Furthermore, maintaining data privacy and security while operating on a decentralised network demands innovative solutions to protect sensitive information.
Decentralised AI: A fair way to use the technology
Decentralised AI represents a promising shift towards a more equitable and fairer utilisation of advanced technology. By distributing the power and control over AI systems among a network of participants, it mitigates the risks associated with monopolisation and centralised authority.
Decentralised AI aligns with the principles of openness, data ownership, and ethical AI practices, making it a compelling avenue for realising the full potential of artificial intelligence while upholding fairness and inclusivity.
Speaking on the aspect of ethics, Dr. Tasca adds:
“It is very important that we advance at the same level not only on the tech side, but also its ethical use. We should strive harder to bring some awareness within communities of users and policy makers to make it fair for use.”