You may have caught glimpses of the recent Saturday Night Live episode, featuring Timothée Chalamet and other cast members sporting similar attire, grooving in vibrant red shorts. The sketch humorously portrayed Troye Sivan as “an Australian YouTube twink turned indie pop star and model turned HBO actor being portrayed by an American actor struggling with an Australian accent.” Sivan’s presence, along with his Sleep Demons Chalamet crew, seemed to have left a lasting impression on many. However, this particular Sivan portrayal was arguably the least spine-chilling aspect of the week.
DeepMind, a branch of Google, introduced Lyria, touted as their “most advanced AI audio technology model to date,” along with two innovative music-generating experiments last Thursday. One of these AI tools allows users to transform a mere murmur into an electrifying guitar riff or a majestic orchestral keyboard solo. The second experiment, known as Dream Track, empowers users to swiftly create 30-second YouTube Shorts using AI-generated voices and musical styles resembling renowned artists like T-Pain, Sia, and Demi Lovato, including Sivan. By selecting an artist and inputting a theme, individuals can have the tool craft lyrics, compose the backing track, and sing the song in the chosen artist’s distinctive style. The level of creativity facilitated by these tools is truly mind-boggling.
The essence of true creativity should not be easily replicable; it should present a challenge. This ease of creation is more unsettling to me than a horde of faux Troy Sivans infiltrating my dreams. The essence of greatness lies in the arduous journey, akin to the character Jimmy Dugan in “A League of Their Own.” If creativity were effortless, it would lose its allure. Charli XCX’s songs resonate because they exude her unique attitude, a quality that shines through even when she pens tracks for others, such as Icona Pop’s “I Love It,” which exudes a sense of fun. It almost feels comical to task a machine with crafting a song about fishing. As the saying goes, “ChatGPT doesn’t possess childhood trauma,” a testament to the challenges faced during the Hollywood writers’ strike.
While these AI devices are not devoid of purpose, they primarily serve to foster idea generation and, in the case of Dream Track, explore novel ways for artists to engage with their audience. Rather than aiming for conventional chart-topping hits, the focus shifts towards creating fresh auditory experiences for platforms like YouTube. Lovato emphasizes how AI is revolutionizing the creative process for musicians, emphasizing the importance of actively shaping the future landscape of the industry.
Google’s latest foray into AI audio technology arrives amidst a complex landscape. Generative AI poses challenges in terms of rights, creating a digital maze of sorts. YouTube, a subsidiary of Google, grapples with an influx of AI-generated music and the intricacies of compensating artists whose work is featured on the platform through label partnerships. Instances like the viral AI-generated track “Heart on My Sleeve,” attributed to “Drake” and “The Weeknd,” led to its removal from various streaming services following objections from the artists’ label, Universal Music Group.
Navigating the realm of AI-generated music presents a conundrum for artists. Even if, hypothetically, the manager of Johnny Cash’s estate does not oppose AI-generated covers of “Barbie Girl,” decisions must be made regarding the utilization of AI tools trained on their work, the development of proprietary tools (as seen with artists like Holly Herndon and Grimes), or opting for a hands-off approach. This dilemma seems to pervade the creative sphere, with designers contemplating their stance on the matter.
The recent discourse surrounding the implications of AI-generated music on intellectual property rights elicited a thought-provoking response. Ed Newton-Rex, head of Stability AI’s audio team, announced his departure from the company, citing disagreements on the notion of training generative AI models on copyrighted material as ‘fair use.’ This departure underscores the contentious nature of utilizing copyrighted works to train AI models, raising pertinent questions about copyright laws’ efficacy in the age of artificial intelligence and its implications for future AI advancements.
In conclusion, the evolving landscape of AI-generated music prompts reflections on authenticity and creativity. While tools like DeepMind’s Dream Track offer a whimsical avenue for creating music akin to established artists, they may fall short of satisfying the discerning tastes of dedicated fans. The narrative harkens back to Nipper, the iconic dog in the RCA logo, listening intently to a phonograph. Just as technology has advanced sound reproduction over the years, AI strives to capture the essence of creativity in real-time moments, albeit with some arguing that digital formats will never match the warmth of vinyl. AI may excel in certain aspects, yet the emotional connection of hearing a beloved artist’s authentic voice remains unparalleled.
