Our COO, @JanetAdamsAI, took the stage at Agents Day in Singapore to discuss the evolving role of AI agents as on-chain decision-makers, joining industry experts Clara Tsao from Filecoin Foundation, Shaw Walters from Eliza Labs, Stefano Bury from Virtuals Protocol, Yurii Kyparus from Wallchain, and Adam Wozney from Akash Network. Janet opened by defining a decision as the act of choosing one action after considering multiple possibilities. This, she explained, distinguishes genuine decision-making from the execution of smart contracts, which follow predetermined if-then logic without evaluation. On-chain decision-making remains in early stages and requires major AI and blockchain infrastructure upgrades before reaching its full potential, she said. The technology's promise lies in eliminating human cognitive biases. She pointed to research showing that parole decisions vary significantly depending on whether judges have eaten lunch, a clear example of how arbitrary factors can distort human judgment. “The power to be more ethical and make better decisions than humans are currently making is huge,” Janet said. This promise remains unrealized in the context of the current paradigm of building larger-scale AI systems, where dominant tech companies control development for profit and the top one percent of earners, perpetuating structural inequalities. Janet posed the urgent question facing the decentralized AI community: how quickly can we transition to AI decision-making supported by diverse inputs, diverse advocates, and the needs of the broadest possible group of people? “It's not a winner-takes-all; it's a community effort,” she said, highlighting that decentralized development must lift everyone instead of prolonging existing power and financial imbalances. Achieving this vision requires overcoming fundamental technical barriers in on-chain AI. Janet highlighted two major challenges: today’s AI systems are too resource-intensive for blockchain deployment, and existing smart contract languages like Solidity lack the flexibility for complex reasoning and the capability to support AI scripting, secure decentralized execution, and interoperability with existing AI and smart contract code. She advocated for neural-symbolic AI, which “works with knowledge graphs and provides a more compact representation of data,” enabling multi-step complex reasoning and offering a more resource-efficient approach for on-chain decisions than LLMs. SingularityNET is addressing these challenges through the Artificial Superintelligence (ASI) Alliance's development of ASI:Chain, a MeTTa-powered Layer 1 infrastructure purpose-built for advanced decentralized AI applications and human-AI agent collaboration. “That is when we're going to start to see the decentralized world that we’re all so passionate about here in blockchain really starting to make inroads into some of the major traditional industries,” Janet said, highlighting the goal of making complex process automation accessible to all humanity.