Blockchain, a technology that has been spoken about for many years, could be used to prevent bias in the data that artificial intelligence models are being trained on, CNBC reports. This concern arises because the data used for training AI models could contain biases or misinformation, thereby leading to incorrect and biased answers. The technology is said to allow developers of AI systems to keep track of the data that the model has been trained on.
Casper Labs, a business-focused blockchain firm, recently partnered with IBM to create a system wherein AI training data can be put on the blockchain. This allows the developers to monitor the datasets, providing a proof of how the AI has been trained. As Medha Parlika, the Chief Technology Officer and co-founder of Casper Labs, explained, the datasets are checkpointed and stored on the blockchain, so that if the AI starts to “hallucinate,” developers can rollback the AI to a previous version. This application of blockchain to AI training data has been described as a “killer use case” for the technology by Sheila Warren, the CEO of the Crypto Council for Innovation.
Warren explained that she believes the verification and checks and balances within an AI system will be blockchain-driven and blockchain-backed. The potential of blockchain to prevent bias in AI systems through keeping track of the training data has generated great interest, with various industries exploring its potential use beyond cryptocurrency. This development highlights how blockchain technology could revolutionize AI, providing a significant advancement in the way AI systems are developed, monitored, and corrected.