by: Jill Koly
There are aspects of Generative AI like ChatGPT that I can’t wait to incorporate into my tasks, and there are some that keep me up at night.
Very quickly it seems ChatGPT went from something that students used to write essays for them to something big and scary that would soon take over the world. With so many big and scary things in the news these last few years, it’s hard to get worked up about yet another one. However, as market researchers we should be keeping an eye on Generative AI. It’s used to create undertested systems that are programmed to sound very convincing. ChatGPT has already been the target of a data breach and at one point was banned from using personal information in Italy.
The Market Research industry in particular should be wary of AI like ChatGPT. It could be used to introduce fraud to our results, or even replace our research with artificially generated results. However, we could also think of AI technology as a tool to enhance our work, and not a replacement for our subject matter expertise.
Generative AI is capable of sifting through large amounts of data very quickly. It can be incredibly useful for getting ideas, creating outlines, drafting emails, and leading you to look at things from a different angle. The only problem is you can’t be sure that what it tells you is true. Just because a response is generated does not mean it is accurate. You need to have a good understanding of the question you are asking in order to properly evaluate the response you get.
Of special concern to Greenwald Research is fraudsters using Generative AI to get even better at creating fake survey responses. This is one reason out of many why it’s so important to have a comprehensive data quality program for online surveys. Is your survey so sensitive that you need to absolutely make sure you don’t have generated responses? You might want to consider what it would take to make your online survey a mail survey.
When you sign up for ChatGPT, it tells you not to share anything sensitive. Why? Because it can’t really say that it will keep your private data private. It’s a nightmare for those of us who work in data privacy, as it should be for anyone with data on the internet (i.e., all people everywhere). Generative AI absolutely should never be used for anything sensitive.
Generative AI offers tremendous benefits. It’s a great tool for efficiency and idea generation. However, the risks of survey fraud and privacy issues should be enough for all of us to weigh those benefits against a large dose of caution.