It might not be the first quality you think of when contemplating the abilities of AI, but resolving wagers has emerged as a viable use case for the technology. These wagers can relate to virtually anything in the real world, with recent examples including ‘What will Kamala say during the North Carolina rally?’
Although AI-powered oracles enable prediction platforms like Polymarket to settle bets, the fallibility of such tools is a topic of debate. This debate ignited earlier this year when Polymarket refuted one of its oracle UMA’s determinations.
From disputed resolutions to bets graded ‘unanswerable’ despite clear evidence, limitations of AI oracles are starting to come to the fore. At the same time, innovative solutions like GenLayer – developed by research lab YeagerAI – purport to address these shortcomings, with the goal of ushering in a new era of accuracy, efficiency, and trust in decentralized decision-making.
The intersection of blockchain and AI
Albert Castella, Co-Founder and CEO of YeagerAI, believes that AI-powered smart contracts could be the answer to resolving contested verdicts in prediction markets. GenLayer is a blockchain platform designed to execute AI-powered smart contracts based on a unique consensus algorithm known as Optimistic Democracy.
“GenLayer’s AI-powered smart contracts could improve the accuracy and reliability of judgments in prediction markets through a more transparent and structured resolution process,” says Castella, explaining that its system of Intelligent Contracts – which combine deterministic code with natural language – can reduce ambiguity and the potential for subjective interpretation.
YeagerAI’s approach aims to tackle several key limitations of existing oracle systems, offering faster resolution times, lower operational costs, improved scalability, and an auditable on-chain decision trail. In essence, what this means is that GenLayer can handle a larger volume of decisions more consistently than human arbitrators, potentially transforming the efficiency of prediction markets.
The phrase ‘AI-powered decisions’ may send a shiver down some spines – particularly those who fear an impending dystopia in which humans are subjugated by robot overlords. But what technologists are seeking to do with AI, in this context, is use the technology to reach definitive conclusions based on verifiable data.
In cases like the Justin Bieber baby gender market on Polymarket, which was controversially graded as unanswerable by UMA before later resolving to Yes, Castella suggests that GenLayer’s approach could have led to a different outcome.
“Our intelligent oracle could have a clear, open process for submitting evidence from validated sources,” he explains, detailing how the system would define specific criteria for acceptable evidence and utilize AI validators to assess this evidence against predefined criteria.
Validators play a key role in GenLayer, collaborating and verifying one another’s work through Optimistic Democracy, which enables them to reach resolutions on non-deterministic instructions. UMA, the decentralized truth machine favored by Polymarket, also uses a form of validation in that all statements can be disputed – with rewards flowing into the hands of those who dispute successfully.
“We use multiple validators with different LLMs, and there is an appeal process that can escalate to the entire validator set for contentious cases,” Castella explains. “Our incentive structure and on-chain transparency further reinforce the system’s integrity.”
Ensuring integrity and preventing manipulation
One of the key concerns in any decentralized system is the potential for manipulation, and prediction markets are no different: an oft-made complaint is that well-capitalized participants can use their influence to ensure the chips fall in their direction. As Castella tells it, GenLayer has no intention of becoming a cautionary tale and has implemented several safeguards to address the issue.
“Our platform prevents manipulation through several safeguards including a diverse validator set, staking and slashing mechanisms, and a 51% honesty assumption,” he shares. “The appeal process and on-chain transparency make it difficult for coordinated groups to consistently manipulate outcomes, with economic incentives aligned to reward honest behavior among validators and Intelligent Contracts constraining the degrees of freedom for validators which are powered by AI.”
Looking ahead, Castella believes AI-powered oracles will continue to have a major impact on prediction markets, transforming the sector by “improving liquidity through faster resolutions and reducing operational costs, enabling more diverse and complex markets.”
The ability of markets to handle real-time data and integrate with other DeFi protocols also opens up possibilities for more sophisticated financial products and other tools. After all, just the other week Ethereum co-founder Vitalik Buterin suggested AI prediction markets could speed up the production of Community Notes on X.
Featured image credit: Rodion Kutsaiev / Unsplash
