Dr Tobias Hann
CEO, MOSTLY AI
The ability to change real data into AI-generated synthetic data will be advantageous for any organisation dealing with privacy-sensitive information.
Synthetic data looks set to transform the way organisations handle privacy-sensitive datasets, enabling them to fully utilise statistical information in a fraction of the time — allowing innovation to flourish quickly.
A data revolution
For sectors dealing with sensitive data that includes Personal Identifiable Information (PII) — particularly the banking and insurance industries — synthetic data will be a game-changer.
Dr Tobias Hann, Chief Executive Officer for MOSTLY AI, says: “Synthetic data is a new, developing category within technology. We think it can change how companies work with data.” He explains that synthetic data is made-up data. The process uses machine learning algorithms that are trained on the original data, so synthetic data represents real data in that all the statistical properties are retained — but without any of the PII.
“Essentially, it’s anonymous data that can be used more easily for sharing both internally and externally, as it’s GDPR-compliant,” says Hann.
In the past, it could take up to six months to access data due to internal compliance processes — that is now reduced to minutes.
Synthetic data benefits
It enables businesses to work with data in ways that weren’t possible before, such as sharing datasets with far more employees, external research partners and vendors. Another major benefit is that time-to-data is significantly reduced. In the past, it could take up to six months to access data due to internal compliance processes — that is now reduced to minutes.
Hann says: “Traditional ways of anonymising data involve manual processes which take time and human resources. Here, we have a simple solution that allows for speedy data anonymisation with little human intervention.”
Shaping the future
MOSTLY AI, established five years ago, is a software vendor that created a pioneering software platform allowing organisations to generate synthetic data based on their existing datasets, at the touch of a button. Launching next year will be the added capability of programmable data which will allow users to change the data, for example, by modifying it to correct biased and unfair information.
Hann concludes: “In the past, we’d say synthetic data is as good as real data. But now, we say it’s better because it can be shaped and formed to become more relevant for our clients.”