Have you got the AI Factor? Robots can now identify a surefire hit… meaning music moguls like Simon Cowell may have to watch out!
- AI can now identify a hit song with almost 100% accuracy, a study has found
Music moguls like Simon Cowell may find their days are numbered.
Artificial intelligence can now identify a surefire hit with almost 100 per cent accuracy, a study has found.
Simply asking people which songs they liked best is a terrible way to identify a hit, researchers found, after recruiting 33 volunteers to listen to 24 recent songs.
But interpreting their brain signals, then getting AI to interpret the results, can distinguish a hit song or a flop with 97.2 per cent accuracy.
The AI can spot a hit with an impressive 82 per cent accuracy after hearing only one minute of the song.
Music moguls like Simon Cowell may find their days are numbered as artificial intelligence can now identify a surefire hit with almost 100 per cent accuracy, a study has found
Its results were checked against the modern definition of a hit – whether a song was listened to on internet streaming services more than 700,000 times.
That is incredibly useful when 168,000 songs are released every single week worldwide, and less than four per cent of those become hits.
The researchers believe the technology could be used to predict which films, TV shows and even social media posts will take off.
Professor Paul Zak, senior author of the study, from Claremont Graduate University, said: ‘By applying machine learning to neurophysiologic data, we could almost perfectly identify hit songs.
‘That the neural activity of 33 people can predict if millions of others listened to new songs is quite amazing.
‘Nothing close to this accuracy has ever been shown before.’
The volunteers in the study listened to songs from a range of genres, including rock music and hip-hop, which had been released in the previous six months.
They wore a simple fitness tracker which recorded their heart rate.
Years of previous research suggest subtle patterns in someone’s heart rate, based on fluctuations in speed or the gap between beats, can show if they are producing the brain chemical dopamine, as they focus on a song, and the hormone oxytocin, as they emotionally connect with it.
The researchers used this to work out when people were in a state of ‘immersion’ and invested in the song, and when they entered a state of ‘retreat’ where they were less interested and only half-listening.
The volunteers in the study listened to songs from a range of genres, including rock music and hip-hop, which had been released in the previous six months (file photo)
This information, analysed using old-fashioned statistics, could identify hits and flops with 69 per cent accuracy.
But that rose to 97.2 per cent when using AI, which was able to calculate immersion and retreat almost 800 different ways.
The people on the study were also asked a series of questions about how much they liked each song.
But how much people said they liked songs they had not heard before was unrelated to whether those songs were hits, researchers found.
Professor Zak said: ‘You don’t necessarily know if a song is going to be a hit, but your brain does, which is crazy cool.’
Although the study used a small sample of songs, Professor Zak said: ‘In the future the right entertainment could be sent to audiences based on their neurophysiology. ‘Instead of being offered hundreds of choices, they might be given just two or three, making it easier and faster for them to choose music that they will enjoy.’
The study is published in the journal Frontiers of Artificial Intelligence.