The World Economic Forum in Davos this year has brought to light some exciting potential for the combination of blockchain and artificial intelligence. Executives are touting blockchain as a possible solution to the problem of bias and misinformation in AI training data. This innovative approach could revolutionize how AI models are trained and used, ensuring that the information they provide is free from biases and false information.
The partnership between Casper Labs and IBM to develop a system for storing AI training data on the blockchain is a promising step forward in this direction. Casper Labs’ chief technology officer, Medha Parlika, explained that by checkpointing and storing training data on the blockchain, developers can have a clear record of how the AI is trained. This means that if the AI system starts to give out false information or “hallucinate,” developers can roll back the AI to a previous, more accurate version.
Sheila Warren, CEO of the Crypto Council for Innovation, believes that using blockchain to verify and monitor AI systems could be the “killer use case” for the technology. This bold statement underscores the potential impact of this new approach, suggesting that blockchain’s immutable and tamper-proof nature makes it an ideal tool for ensuring the accuracy and integrity of AI systems. As the discussion at Davos has shown, the pairing of blockchain and AI has the potential to revolutionize the way we think about data integrity and the development of AI technologies.