Jun 20 2023
Using AI for Neuroforecasting
I’ve been following AI (artificial intelligence) news very closely, including all the controversies and concerns. I tend to fall on the side of – AI is a powerful tool, we should continue to develop it and use it responsibly. We don’t need to panic, and highly restrictive laws are likely unnecessary and counterproductive. But there are legitimate concerns about the power of AI, especially in the “wrong” hands. I also think the greatest disruption to our lives might not come from cyberterrorists (although a legit concern) or AI run amok, but from marketing. Giving companies who see us only as customers the power to predict our every move gives me pause.
This AI news item falls into this latter category – the use of machine learning AI to predict which songs people will like. Seems innocuous, but I think it furthers a trend that has some serious downsides. This is what the researchers did:
Traditionally, song elements have been measured from large databases to identify the lyrical aspects of hits. We took a different methodological approach, measuring neurophysiologic responses to a set of songs provided by a streaming music service that identified hits and flops. We compared several statistical approaches to examine the predictive accuracy of each technique. A linear statistical model using two neural measures identified hits with 69% accuracy. Then, we created a synthetic set data and applied ensemble machine learning to capture inherent non-linearities in neural data. This model classified hit songs with 97% accuracy.
This kind of approach is called neuroforecasting – predicting people’s likes and dislikes based upon their brain activity and physiological responses (like a lie detector but for your reaction to music). First let me point out that this study used a synthetic set of data, and is therefore just a proof of concept – this approach can theoretically work. They need to test this in the real world, and see if it can predict hits, not just match the model to existing hits. But let’s assume it works, and the 97% accuracy hold up. What will this mean for the music and streaming industries?