During a panel session at SmartCon 2023, Eric Schmidt, former CEO of Google, and Sergey Nazarov, co-founder of Chainlink, engaged in a fascinating conversation that explored the intersection of blockchain and artificial intelligence (AI).
The discussion delved into the latest developments in AI, emphasizing the significance of blockchain in ensuring data security and fostering trust. Additionally, the participants raised concerns about the potential drawbacks associated with generative AI systems, like large language models (LLMs).
Eric discussed the major advancements in AI that have taken place in the past year, specifically highlighting the widespread adoption of LLMs. He noted that substantial investments, often reaching hundreds of millions of dollars, are now being made to train these models—a significant departure from the past when software development was more cost-effective.
Furthermore, Eric acknowledged the transformative potential of AI in sectors such as healthcare and education, while also alluding to the need for caution due to associated risks.
As evidence, he mentioned the evolution of OpenAI’s ChatGPT, the development of Anthropic Claude 2, and even Nvidia’s contribution in making AI chips more affordable compared to previous years.
AI and blockchain
Sergey spoke about the increasing role of AI in smart contracts and blockchain technology, with a particular emphasis on the utilization of Chainlink functions to provide computational resources for trading strategies and more.
The Google co-founder added that there is a general-purpose computational power inherent in smart contracts. He suggested that the use of AI in the blockchain context is a natural evolution and that the two aspects of today’s technological revolution – LLMs or blockchains – have their merits.
He went further to give a historical perspective, comparing the evolution of AI and blockchain technology to industrialization processes. He highlighted the rapid evolution of software and its potential to transform the industry at an accelerated pace.
Concurring with Eric, the Chainlink co-founder said that there is a symbiotic relationship between blockchains and LLMs. He added that blockchains can define key parameters for LLMs, while LLMs perform specific computations that blockchains or Oracle networks can’t. The result of these two dimensions of life today is a foundation for an unpredictable world that will be driven by technology.
More attacks and challenges will make the industry better
Eric emphasized that facing challenges and obstacles is an essential component for the growth and enhancement of infrastructures like Chainlink. He believes that overcoming these challenges will drive innovation, improvement, and resilience in such systems.
Looking back again at Gmail, and the relationship between the level of security in a system and the number of potential points that attackers can target, he said that the more ways or points within a system that can potentially be targeted by attackers (more attack surfaces), the greater the opportunities there are to enhance the system’s security.
He compared the security of a corporation’s email servers to that of Gmail. The argument is that Gmail is more secure because it is under constant attack (100 times a second), which has allowed it to learn from and defend against a wide range of attacks. This constant exposure to threats has made Gmail more secure over time.
The level of security in a system depends on whether the system is heavily used and constantly attacked. In the case of a decentralized model or a centralized system, if they are both heavily used and frequently targeted, they are more likely to be secure because they have to withstand the attacks.
What the industry should focus on
Speaking about the future of a safer technology space, Eric said that it’s important to focus on gaining control over one’s identity in a trusted manner. He added that those working on similar technologies should find ways to integrate safe identity and security systems with the Chainlink infrastructure.
The speaker expressed worries about the potential consequences of not having a system in place to manage the information created by AI, particularly when it involves generating fake content or impersonating human identities. “Without it, how are we going to run life? How are you going to know what somebody says is true.”
Eric raised a critical question about how we can ascertain the truth and trustworthiness of information and individuals in an increasingly digital world. This is particularly crucial for trusted payments and other online transactions.
He added that the credibility of the entire system can be compromised if misinformation and a lack of trust become pervasive as the internet has become an integral part of modern society, making it impossible to simply “turn it off” to avoid problems.
Democratization of AI
As the conversation continued, Sergey questioned the decentralization of AI. He pondered whether AI might be decentralized, with networks of individuals collaborating to run AI systems, or if it would remain under the control of a few dominant companies due to their access to data and resources.
Eric responded, saying he foresees open-source AI systems being accessible to the vast majority, providing powerful solutions for various applications. However, at the frontier of AI development, it remained uncertain whether one or two giant companies with specialized algorithms would dominate, or if a consortium of smaller, agile teams could compete.
Reflecting on his journey at Google and how they competed with other huge companies, he noted that there has always been a fight between open and closed systems and acknowledged that it was a fundamental question in the tech world. He said that the current race to control the future of intelligence is one of the most significant battles in the world.
