Between Two Joels: Synthetic Data Edition

The experts have spoken...

Play Video

It turns out the hype around synthetic data is both real and exaggerated. During our Q&A session on Wednesday, our AI Experts, the Two Joels along with co-founder Ian Ash discussed what's we should and should not be doing as it relates to synthetic data. We talked about things like:

1. Can this methodology be used in B2B?

2. How do you correct for bias?

3. How does it relate to the Digital Twin Universe? 

Some key insights from the session:

•  Synthetic data is revolutionizing data collection by mimicking real data in situations where it's unavailable or costly to obtain. It offers exciting possibilities for some situations, but doesn't do the job well in others (think brand tracking vs concept testing).

• Our AI experts created an analytical framework to leverage when considering how to use synthetic data. The framework resembles a pull manufacturing model, emphasizing the importance of analyzing synthetic data at the methodological level. By delving into the methodology behind synthetic data generation and analysis, we can gain deeper insights into its validity and reliability. 

• Overall, there's a general lack of research on the topic. There is also a lack of standards on what it means to have accurate and reliable synthetic data. Before we can start validating synthetic data's validity, it's important to lay the groundwork on when it is useful and when it isn't.