Tech Talk: The Future of Artificial Intelligence is Web3

After sharing a variety of real-world applications, Gary gave a brief explainer of Web3. He introduced this segment with examples of how the internet has evolved:

• Web 1.0 – one-way sharing of information by putting it “out there” for other people to access with static read-only web pages
• Web 2.0 – interactive data sharing, such as on social media
• Web 3.0 – decentralized data, meaning we take back control of our own data on our device.

Gary emphasized that the core principle of Web3 is being able to have a digital identity that can’t be erased. Blockchain is a key part of making this possible. Essentially, it acts as a tamper-proof digital ledger to record information.

The term Web3 may not be as well known as artificial intelligence or blockchain, but there are many technologies built using Web3 principles, such as:

• NFTs
• Metaverse
• Decentralized Finance (DeFi)
• Smart Contracts
• Bitcoin
• Digital currency
• Tokenization

After this introduction to Web3, Gary transitioned into the “deep cut” of the presentation: how Web3 can help keep AI in check. He emphasized that we humans are in charge of AI. It’s up to us to put the right controls in place. Web3 and blockchain give us a variety of tools to implement checks and balances that will help ensure the use of AI-generated data is done in a transparent and secure manner.

Explainable AI (XAI) is an important development to help build trustworthy AI systems. Gary compared it to doing an autopsy on AI. XAI keeps humans “in the loop” by helping us understand how the AI arrives at a particular answer. This capability is important to establish trust in the AI, especially as people start making increasingly important decisions based on AI-generated data.

Gary acknowledged there are some drawbacks to XAI. Currently, many of the XAI models take longer to generate answers, which can be prohibitive for solutions where time is of the essence. There are also challenges with protecting the privacy of source data, since this method pulls back the curtain to show what source data was used to generate a given answer. However, the biggest hurdle to overcome is simply explaining the complex AI processes in a way that is complete and accurate but still understandable to the average person.

Gary emphasized that AI may be a hot topic right now, but the market for XAI is going to be massive. These systems will be able to show logs, search, reports, and workflow for everything generated by the AI. When tethered to blockchain, this information becomes tamper-proof. When paired with Web3, every AI process can be linked to the digital identity of the person who initiated it, adding accountability and reducing the potential for fraud. Use of biometric authentication, such as using a person’s unique heartbeat rhythm to verify their identity, adds another layer of security while also providing proof of humanity. These types of technologies are just a few of the ways that we can implement checks and balances on AI-generated content.

Separately, AI, Web3, and blockchain each have their own challenges and risks. Together, they create a powerful synergy that has the potential to avoid many of the worst-case scenarios of AI gone wild. Ultimately, it’s up to us to put these systems and protocols in place.

Project Examples

Gary shared examples of projects that are using AI, blockchain, and Web3 technologies to help verify data and provide authentication capabilities:

AI Watermarking
A variety of methods can be used to add a watermark to AI outputs. A couple examples:
Blockstamped – Blockchain used to “watermark” news articles to help combat fake stories
Sony, Nikon and Canon Partner to Fight Deepfakes

CGI TrustedFabric
TrustedFabric is a solution for sharing externally while maintaining control of the data
One example of TrustedFabric in action is a project with the European Space Agency to create a cloud-based Universal Virtual Flight Data Recorder. This would become the primary method with the physical “black box” as a backup.

CGI and Google Cloud have partnered on the use of PulseAI
This generative AI solution consistently achieves processing accuracy rates that exceed 90%, increasing operational efficiencies and accelerating the launch of new business processes.

CGI Trade360 is used with Skuchain’s blockchain platform
This project demonstrates how trade finance platforms can be integrated with blockchain to streamline processes and significantly reduce transaction time.

Immutable Vaccine Supply Chain (Monkeypox)
This proof of concept model enables universities to collaborate with WHO, CDC, and vaccine providers to protect the personally identifiable information (PII) of the people who need treatment.

VIZER Vault
This project with the University of Tennessee and CGI created a health information digital vault that allows people to keep all their health records in one place and give consent for who gets that information.

Decentralized Autonomous Organization (DAO)
This new form of company allows for a decentralized decision-making structure governed by voting shares and smart contracts. Tennessee is one of only three states where DAOs are legal. Only five other countries besides the US recognize DAOs.

Heartbeat Authentication
CGI and it’s partners are working on solutions that would link execution of an AI process to a digital identity, verified with the person’s heartbeat. This type of authentication could also be used for a “kill switch” on AI systems. The audience was particularly interested in this project, with several questions about it during Q&A. One of the main concerns was what happens if a person’s heartbeat is altered, either by AFIB, while having a heart attack or even after having a heart transplant. Gary assured us that there are failsafes and solutions in place for those types of scenarios.

Hosted In Partnership With