• John Ennis

Eye on AI - April 30th, 2021

Welcome to Aigora's "Eye on AI" series, where we round up exciting news at the intersection of consumer science and artificial intelligence!

This week, we’ll be looking at ways AI helps eliminate biases in consumer surveys, with particular focus on a recent text-to-speech study that seeks to answer the question of whether or not text-to-speech AI can limit interviewer biases.


Enjoy!


New Study Reveals How Text-to-Speech AI Limits Interviewer Bias



Speech hints at more than simply the words spoken. Tone suggests intent. Accent alters perception. Innumerable auditory variances change listener perceptions in known and unknown ways. That’s a big problem if you're a marketer looking to conduct an unbiased customer survey. Recognizing these inherent biases, researchers at EyeQuestion conducted a study to see whether or not those biases were still present using text-to-voice AI (i.e. a smart speaker read pre-written survey questions to participants rather than a human).


The study worked like this: two blind taste test surveys were conducted, one using a human and one using text-to-voice, in which panelists tasted the same Fruit-tella product twice but were told one was a healthier, sweeter version of the original product, and were then asked to rate each based on six different attributes. Results were combined to calculate each participant’s overall satisfaction score, which showed a significant difference in scores between the two surveys.


“In the interview with a human interviewer, panellists rated the new variant significantly higher than in the survey where Text-to-Speech and Speech-to-Text techniques were used,” writes the EyeQuestion research team. “This was not the case for the regular variant. Within the human interviewer session, panellists had a significant preference for the new variant. In the survey where Text-to-Speech and Speech-to-Text techniques were used, no preference was found.”

While we shouldn’t jump to too many conclusions from a single study, these results do show that marketing researchers can use text-to-speech and speech-to-text techniques to limit the interviewer bias in their results.


AI Is Changing Market Research for the Better



The previous study’s results build on an idea we’ve been following for some time, which is that AI helps eliminate biases in consumer studies. You may remember a study we highlighted in January that used VR to reveal light’s effect on taste perception, or our post in October of last year that focused on how VR changed people’s taste perception based on setting. Even still, some market researchers continue to rely on in-person surveys using traditional methods that are prone to more biased results, which often lead to countless marketing mistakes that may have been avoided.


As studies continue to reveal how traditional survey methods often lead to compromised results, and as people continue to socially distance in the wake of the pandemic, traditional in-person survey methods are becoming less tenable, not to mention expensive. Thankfully, as voice AI continues to evolve into the post-pandemic standard, newer, better methods are emerging for market researchers looking to conduct less-biased consumer surveys with the help of VR, smart speakers, and voice-to-text / text-to-voice AI. The marketing survey landscape is changing, and in my view, it’s for the better.


Other News



That's it for now. If you'd like to receive email updates from Aigora, including weekly video recaps of our blog activity, click on the button below to join our email list. Thanks for stopping by!