Replica Analytics - An Aetion Company

The dominant narrative today about generative AI is that this technology will amplify privacy risks. Different scenarios have been posited, ranging from training machine learning models on personal information which the models then memorize, to users sending personal information to interactive generative AI models. While these are realistic risks, generative AI can also be a tool that enhances the privacy of data. In fact, generative AI has been applied quite extensively in the health data space for the last few years as a privacy enhancing technology by generating digital twins.

This talk is an extended version of one that was presented recently at the International Institute of Communications / Canada annual conference. The presentation is targeted at senior and executive level participants who want to learn about how generative AI is used as a privacy enhancing technology by generating digital twins of patients. The digital twins retain the properties of the original patient data but they are not the same as the original patient data. This is the idea behind synthetic data. The talk will summarize how this type if data is generated and the existing evidence on its privacy and utility.