Spotify, on Thursday, strengthened its partnership with Google Cloud, employing large language models (LLMs) to refine the accuracy in identifying users' listening patterns across podcasts and audiobooks, with the objective of providing more personalized recommendations.
Large language models, including OpenAI's ChatGPT and Google Bard, leverage artificial intelligence and undergo training on extensive datasets to generate diverse content and text.
Google Cloud, a subsidiary of Alphabet, features several LLMs like PaLM 2, Codey, Imagen, and Chirp, trained on various data types, including text, codes, images, audio, and video.
As an early adopter of AI for music recommendation algorithms a decade ago, Spotify is now extending the application of LLMs beyond music, incorporating them into non-music content such as podcasts and audiobooks.
The Swedish streaming giant is aiming to diversify its revenue streams by expanding its offerings in formats like podcasts and audiobooks, anticipating high-margin returns from these ventures.
Gustav Söderström, Spotify's chief product and technology officer, emphasized the collaborative synergy between Spotify's advancing technology and Google Cloud's commitment to providing an innovative platform for their products. The partnership aims to leverage the emerging capabilities of generative AI.
In addition to delivering personalized recommendations, Spotify is exploring the use of LLMs from Google Cloud to enhance the safety of the listening experience and identify potentially harmful content. This underscores Spotify's dedication to user safety and content moderation.