In a recent fireside chat, Dominic Williams, the visionary founder and Chief Scientist of DFINITY, shared his insights on the evolving landscape of artificial intelligence (AI) and its intersection with blockchain technology. The discussion, hosted by Tracy Trachsler, DFINITY’s Head of Institutional Relations, shed light on the Internet Computer’s (ICP) remarkable achievements and the exciting innovations on the horizon. From the potential of running AI on blockchain to the creation of MotokoPilot, these advancements are set to redefine the future of decentralized applications (dApps) and smart contracts.
GPU (Graphics Processing Unit) Dilemma
GPU (Graphics Processing Unit) is a critical component used to train AI models and run inference tasks. It is essential for processing the large amounts of data required by AI models, such as the large language models (LLMs) like OpenAI’s ChatGPT and Meta’s Llama 3. In the context of decentralized AI models, GPUs can be sourced from Decentralized Physical Infrastructure Networks (DePINs) to provide the necessary computational power.
However, deploying AI models involves a crucial choice between centralized and decentralized hosting. Popular large language models (LLMs) like OpenAI’s ChatGPT and Meta’s Llama 3 run on centralized servers, which, despite their efficiency, are vulnerable to cyberattacks and single points of failure. Dominic Williams advocates hosting AI on the blockchain to realize its full societal potential. Decentralized models, particularly those using GPU resources from Decentralized Physical Infrastructure Networks (DePINs), struggle to match the scale and cost-effectiveness of centralized cloud providers like Amazon Web Services.
The Internet Computer (ICP) offers a unique approach by enabling AI to operate within smart contracts, providing tamper-proof, unstoppable, and autonomous benefits. Unlike most blockchains, ICP’s smart contracts can fully operate AI on-chain thanks to their 32-bit WebAssembly (WASM) virtual machine, which provides 4GB of memory. As ICP transitions to 64-bit WASM with 16GB of memory, it may soon support the on-chain deployment of advanced models like a lighter version of Llama 3.
Real-World Applications: AI-Driven Smart Contracts in Action
The potential of running AI on the blockchain extends to various real-world applications. For instance, a legal firm could employ an AI oracle to analyze client communications, legal filings, and historical case law, drastically reducing the time needed for trial preparation while minimizing the risk of errors. If hosted on centralized servers, such a model would be vulnerable to cyberattacks and outages. However, running it on the Internet Computer’s tamper-proof and unstoppable smart contracts ensures both security and reliability.
Similarly, dApps can leverage AI-driven smart contracts to enhance security and functionality. For example, digital wallets implemented as smart contracts can offer greater security through autonomous operation, preventing any unauthorized modifications to the code. These wallets could also include AI models that execute trades based on predefined conditions, such as automatically converting Bitcoin holdings to stablecoins when certain price thresholds are met.
Another promising use case involves AI models that audit Solidity code on Ethereum, identifying vulnerabilities and establishing trustworthiness. The integrity of such audits relies on the autonomous nature of the smart contracts that contain these AI models.
MotokoPilot is also a pivotal aspect of AI on ICP, it is an AI-powered development tool designed specifically for Internet Computer developers. It uses a comprehensive dataset derived from open-source Motoko code repositories to train state-of-the-art AI models. These models are fine-tuned to assist developers with tasks such as code generation, documentation, debugging, and refactoring.
The future looks promising as ICP prepares to unveil a significant development in the intersection of AI and blockchain. The upcoming beta launch, hinted at by Dominic Williams, is expected to introduce a new dimension to this evolving field, further solidifying the Internet Computer’s position at the forefront of decentralized innovation.
Conclusion
The Internet Computer’s advancements in AI and blockchain are paving the way for a more secure, autonomous, and innovative digital landscape. As these technologies continue to evolve, the possibilities for decentralized applications and smart contracts are boundless, promising a future where AI and blockchain work hand in hand to drive the next wave of technological innovation.