Billie Eilish wasn’t the only big winner at the 62nd Grammys this year. Artificial intelligence scored as well, both on the red carpet and by predicting the winner of Record of the Year.
Billie Eilish and artificial intelligence won at the 62nd annual Grammys, with data science company DataRobot accurately predicting the winner for record of the year. It successfully picked Billie Eilish as the record of the year winner with ‘bad guy’ and it had pegged her as a close second for song of the year, with the title track ‘bad guy’ a category which she ultimately won.
Lizzo garnered the most nominations this year, but Billie Eilish took home the awards for record, song and album of the year, along with Best New Artist. Eilish was the youngest person ever nominated for all four awards at the same time. Lil Nas X was nominated six times and walked away with two wins, according to CBS.
Taylor Larkin, a data scientist with the company, used the DataRobot enterprise AI platform to predict the winners for song and record of the year. DataRobot leverages machine-learning models to learn from the past to predict the future. “For Grammy predictions, the past are previous nominees and winners. The future are the upcoming Grammys,” Larkin said right before the Grammys,.
SEE: How And When To Watch The 2020 GRAMMY Awards (CBS)
“Using historical data about the Grammy nominees and winners dating back to 1959, I gave this information to DataRobot, which created hundreds of candidate machine-learning models. I then selected the model that had the best performance on the five most recent award ceremonies,” he explained.
As for accuracy, Larkin said DataRobot “tests and validates each model, ranking them on a leaderboard, by predictive performance. The final model used to make predictions does about 77% better compared to just randomly predicting the past five winners in either category.”
“The DataRobot platform and the data are constantly evolving,” Larkin added. “The former in terms of accuracy and capabilities, and the latter, in terms of variety and availability.”
To recap, this year’s DataRobot predictions for both song and record demonstrated how close the competition was:
Song of the Year
- Lewis Capaldi, “Someone You Loved” – 35.25%
- Billie Eilish, “bad guy” – 33.53%
- Lana Del Rey, “Norman F***ing Rockwell” – 26.86%
Record of the Year
- Billie Eilish,”bad guy” – 31.10%
- Lil Nas X, “Old Town Road” – 29.53%
- Bon Iver, “Hey, Ma” – 19.23%
SEE: The Grammys go high-tech with IBM and artificial intelligence (TechRepublic)
. The team at IBM took over 18 million documents and data sources on this year’s Grammy nominees and distilled them down to the most interesting nuggets about each of the artists to create a modern day version of the 90s’ “Pop Up Video.”
Grammy Insights with Watson used Watson Discovery, Natural Language Processing (NLP) and Speech to Text technologies to analyze the red carpet audio and video in real time to identify which artist is on the red carpet and the nature of the interview between the artist and the host. Pulling from those millions of data sources, IBM displayed information cards about the artist in real-time with relevant information about the artist and the topic of conversation so that fans watching grammy.com would be able to immediately see the interesting tidbits of information.
The AI-enhanced data began four hours before the televised award show, on Grammy.com, thanks to a partnership between IBM and the Recording Academy. The last hour of Grammy Insights was live on CBS.
The red carpet high-tech extras with IBM wouldn’t be possible without the advancement in NLP, which is the technology that enables AI to understand human language. It’s how AI technology can understand, decipher, summarize and contextualize human language and identify compelling insights.
And as NLP technology advances, so does AI’s ability to assist people, and in the case of the Grammys, improve a music fan’s experience while they watch the show.