How to harness generative AI in music and video production without displacing artists | TC Disrupt24

Generative AI in Music and Video Production at Disrupt

• A year ago, it would have been too early to discuss generative AI in music and video production at Disrupt, but significant progress has been made since then. (22s)
• Mikey, the CEO of a generative music company, visited a songwriting class at Berkeley College of Music, where students used AI tools to create songs with non-traditional music makers, finding the experience emotionally moving and empowering. (42s)
• The approach to introducing AI in music should involve open dialogue and reassurance, highlighting AI as a tool that empowers artists to produce more music and explore new creative avenues. (1m44s)
• Kakul, the CEO of Splice, a leading company in the sample space, explained that their AI-based tools help students and creators enhance their music production skills, and young people are eager to explore and innovate within the industry. (2m24s)

AI in Music Education and Creation

• The use of smartphones in filmmaking has empowered students who lack access to traditional camera equipment, allowing them to create films and explore their creativity, although it has also impacted the demand for professional photographers. (3m46s)
• The discussion highlights the impact of AI tools on creativity, noting that while they can enhance creativity, they may also affect artists' roles. (4m27s)
• The speaker, AMT, mentions their work on visual generative models at Luma and draws parallels to how the iPhone revolutionized photography by making it accessible to everyone, preserving visual memories in camera rolls. (4m41s)
• The evolution of technology is discussed, emphasizing that its true potential is realized when it becomes accessible to everyone, similar to how television and entertainment have become widely available. (5m46s)
• Generative AI is seen as a continuation of this technological arc, enabling people with taste and skills to create extraordinary things, even though some traditional jobs may be displaced. (6m21s)
• The democratization of art creation is highlighted, with the example of how expensive film once limited artistic expression, whereas now, more people can experiment and create art due to accessible technology. (6m57s)
• The appreciation for mediums like photography has increased as they have become more accessible, challenging the notion that their value has diminished over time. (7m53s)
• The widespread availability of technology, such as iPhones, enhances individual appreciation and value for photography, as it allows more people to engage with the medium. However, human skill and taste remain crucial in differentiating exceptional work from the ordinary, even with advanced tools and technologies. (8m13s)

Democratization of Art Creation through Technology

• Generative AI is seen as a continuation of previous technological advancements, where personal use leads to unexpected applications, such as using a phone to remember where a car is parked. The impact of generative AI is expected to be significantly larger than past technologies. (9m23s)
• There is controversy surrounding the training of large language models (LLMs) due to the use of vast amounts of human-created art, often obtained through web scraping. This raises issues about the legality and ethics of using internet content without proper licensing. (9m58s)
• Companies like Splice emphasize the importance of respecting creators by maintaining relationships with artists and using licensed content for training their models. This approach highlights the need to honor the craft and rights of sound designers and musicians. (10m51s)
• The issue of respecting creators' rights is not new, as seen in platforms like Flickr, which emphasized the importance of ensuring users had rights to their publicly shared images. This responsibility continues to be crucial in the context of generative AI and content sharing. (11m36s)

Copyright and Ownership in the Age of AI

• Companies are building AI-based tools, and it's essential to be careful with what is said about this topic, as it's still in its early stages, and there are many unanswered questions, such as copyright issues associated with AI-generated songs (12m10s).
• Users of AI music tools often ask difficult questions about copyright and ownership, and the answers are not always clear, highlighting the need for patience and further exploration of these issues (12m28s).
• Working with artists like Timbaland, who has joined as a strategic adviser, provides valuable insights into how to make a generative music startup that honors the work of artists (13m7s).
• Timbaland is a forward-leaning artist who is open to using new technologies, and his involvement helps to promote the use of AI tools in music production (13m52s).
• The best users of AI music tools are often professional music producers who have skills and taste, and they are able to create high-quality content using these tools (14m6s).
• The use of AI tools in music and video production does not disrespect previous technologies or artists who do not use them, and it's essential to recognize that skills and taste are still necessary to create high-quality content (14m48s).

The Role of Skill and Taste in AI-Generated Content

• The key to balancing creative pursuits with AI tools is to recognize that people with skills and taste will still be able to create content that captures hearts and imaginations, regardless of the technology used (15m26s).
• The difference between a skilled user and an unskilled user of AI tools is not the technology itself, but rather the skill and taste that the user brings to the creative process (15m38s).
• Artists will evolve and adapt to new technologies, and their work will reflect the tools they use, with those who embrace new technologies doing better than those who do not (16m3s).
• Creative people will always find new and innovative ways to use technology, and the best musicians will use whatever tools are available to express themselves (16m45s).

User Base and Applications of Generative AI Tools

• Top creators need to feel secure in their rights to the content they produce, and this is an important consideration for companies developing generative AI tools (17m12s).
• The user base for music production companies can vary, with some companies catering to professionals and others to hobbyists or average people who have never made music before (17m45s).
• Music is a universal part of human culture, and everyone has tastes in music, making it accessible to people from all walks of life (18m1s).
• The user base for video production companies can also be diverse, including professionals, semi-professionals, and people from other industries such as fashion who use the tools for unintended purposes (18m20s).
• Democratizing powerful technology can lead to new and innovative use cases, as people are more likely to experiment and try new things when the cost is low (19m5s).
• The cost of video production can be prohibitively expensive, but making it more accessible and affordable can lead to new opportunities and use cases (19m15s).
• Companies are already using generative AI tools to produce high-quality videos, such as launch videos, at a fraction of the cost of traditional production methods (19m19s).
• There is a growing interest among students, not just film students, in using AI models to visualize ideas and concepts, although the current models are not yet highly effective at this task. The developers are working to improve this capability as they recognize its demand. (19m36s)
• The user base for AI tools includes a significant portion of studios and professionals, accounting for about 34-35% of users. The remaining users are individuals who explore the visual medium for personal reasons, such as therapeutic benefits or personal visualization needs. (19m56s)
• Many users find value in AI tools for visualizing thoughts and ideas, which can be crucial for those who think visually. This capability allows users to share their mental images with others, providing a significant advantage. (20m41s)
• The customer base for AI tools in music production includes both professionals and passionate individuals. An example is a software developer who has been creating music for over 20 years and aims to release at least one musical project annually. He uses AI tools to explore new musical styles, such as transitioning from ambient EDM to a lo-fi hip-hop style, which he found empowering. (21m10s)

Advancements and Future of Generative AI

• The progress in AI technology is rapid, with significant advancements in the past year. The industry has been focused on establishing foundational elements like efficient GPU training and clean data. Training AI models is described as an empirical science, with many models discarded before achieving success. Each iteration improves success rates and expands the capabilities of AI models, such as moving towards whole song generation. (22m21s)
• In 2024, advancements in video production using generative AI are expected to improve significantly, focusing on maintaining logical consistency and causality throughout the video content. This involves ensuring that events in the video are coherent and follow a logical sequence, similar to how sentences relate in a language model. (23m30s)
• The goal of these AI models is to produce content that aligns closely with the creator's intent, moving beyond simple descriptions to more complex narratives, such as depicting a full story arc. This development is anticipated to progress faster than expected, leading to a more engaging and dynamic content creation process. (24m13s)
• The ease and intuitiveness of creating content with generative AI are expected to lead to a significant increase in the amount of available content. This shift will necessitate new economic models and systems for content discovery and recommendation due to the abundance of content. (24m55s)
• As AI models improve, the quality of generated content, such as music and video, will continue to enhance, allowing for highly personalized and tailored content. This transformation will shift the focus from merely searching for content to generating specific content that meets individual needs, making the medium more valuable overall. (25m59s)
• With the rise of generative AI in music and video production, creators can focus on making niche stories that appeal to a smaller, specific audience, rather than trying to create mass-appealing content that caters to the lowest common denominator (26m41s).
• This shift enables artists to produce content that is entertaining and funny to a particular group of people, even if it's not relatable to a broader audience, and new economic models will emerge to support the consumption and sharing of this content (27m11s).

Content Overload and the Rise of Niche Content

• The problem of content overload is not new and was exacerbated by the rise of smartphones, which created an environment where there is more content available than people can consume (27m25s).
• AI accelerates this trend but doesn't change the core premise, and the marketplace dynamic will continue to elevate better content to the top, with "better" content being defined as that which connects with people emotionally, rather than just being high-fidelity (28m5s).
• However, the increased use of AI will also lead to the creation of low-quality, potentially hurtful, or dangerous content, making it essential to develop responsible technology that prioritizes human well-being and creativity (28m20s).
• The development of responsible technology is crucial, and it's the responsibility of creators, technologists, and the audience to prioritize this goal, whether from a rights perspective, a creator perspective, or a focus on making useful technology (28m42s).