Disrupting the Media Industry with Generative AI: Challenges, Questions and Opportunities

Get ready for the staggering growth of the global artificial intelligence (AI) market, which is projected to reach a breathtaking USD 1394.30 billion by 2029. According to Globe Newswire, the AI market is set to surge at a compound annual growth rate (CAGR) of 20.1% from 2022 to 2029, solidifying AI’s expanding influence in the entertainment realm. AI is a powerhouse in the media and entertainment industry, enhancing gameplay realism, detecting fake stories, combating plagiarism, streamlining production schedules, personalizing content, empowering creativity and driving sales and marketing.

Across the media and entertainment world, the generative AI (Gen AI) revolution has sparked excitement and trepidation. This was the focus of a lively roundtable discussion during DataArt’s IT NonStop conference in November 2023, during which industry insiders shared their thoughts, questions, concerns, and expectations.

Excitement About Generative AI in the Media Space

From fuelling artistic innovation to automating intricate tasks, Gen AI captures the imagination of creators and industry leaders alike. In this exploration, we delve into the core of this excitement, unravelling how Gen AI is revolutionizing the media space and the interplay between continuity and disruption it introduces.

Creative Aspects: Using AI Tools to Boost Creativity and Productivity

AI tools are proving to be invaluable for many artists, helping them discover novel concepts and refine their artistic expressions. Using Machine Learning (ML) models, content creators can break through traditional creative barriers, seamlessly generating ideas, visuals, and even entire narratives. Although ML models may assist in generating unexpected variations of pre-existing images, the artist remains irreplaceable in providing these images with artistic context and purpose.

In a recent Oxford study assessing the impact of AI on the arts, findings revealed that most artists showcased an “essentially unchanged” connection to their craft, emphasizing their dedication to addressing human-centric queries over technical concerns. Contrary to the notion of complete process replacement, researchers discovered that artists integrate ML into their practice using five key activities:

1. Technical research.
2. Using and building machine learning models.
3. Using and creating datasets, training models.
4. Combining models.
5. Curating outputs.

For instance, a visual artist and software engineer, Helena Sarin, delves into experimentation with diverse ML models to attain the intended results.

“I usually start from noise. I make small images, and then I pass them through several GANs. That’s called GAN-chaining. Then, I upscale them. The first output of the GAN is basically completely unexpected. Let’s say I give it a bunch of my flower photos, and it generates hopefully something that looks like a flower, but because they are small and I start upscaling, the next GANs can take me to some images that, semi-abstracted, might have nothing to do with flowers. Then, I decide whether to do something with them or abandon the idea and start something different”.

Drawings before application of a trained CycleGAN ‘filter’

Drawings after application of a trained CycleGAN ‘filter’

We witness how Gen AI models significantly impact artists’ creative processes, blending continuity and disruption. This transformation is observed in the restructuring of creative workflows, the re-evaluation of AI-generated outputs, and the evolving embodied experience of artists. The influence of Gen AI is not a fleeting trend but a lasting paradigm shift, fundamentally reshaping creative expression.

Automation of Things

Edward Ginis, co-founder of OpenPlay, and one of the panellists at the IT NonStop conference, claims that an exciting impact of Gen AI lies in creating abundant opportunities for automation. This shift envisions humans being out of the processing loop, as exemplified in the music sector, where Gen AI automates tasks like translating music metadata into multiple languages and suggesting personalized playlists based on vast data analysis. Leading platforms such as Spotify and Apple Music already leverage AI to enhance user experiences and recommending songs tailored to individual preferences. This broadens listeners’ musical horizons and empowers artists to connect with new and engaged audiences.

In addition, Gen AI, exemplified by tools like Amper Music’s Songwriter and AIVA, extends its influence on songwriting, enabling musicians to generate melodies, chord progressions, and lyrics effortlessly. Moreover, with AI assistance, arrangement tools like BandLab’s Band-in-a-Box and Presonus’ Notion enhance song structure.

As the recording process undergoes automation, particularly in mixing, artists gain more time to focus on the core creative process rather than getting bogged down by technicalities. This transformative capability proves invaluable, especially in the visual-oriented platforms driving music consumption. Recognizing the significance of delivering visually engaging content, artists often face resource constraints. Gen AI tools now provide a solution, enabling artists to generate visual components automatically while concentrating on the recording process. This convergence of automation and creativity not only defines the media industry’s future but also empowers artists with unprecedented tools to elevate their craft interactively and dynamically.The Evolving Role of AI Algorithms
Machine learning algorithms play a pivotal role at the core of AI-generated content. These algorithms meticulously analyze vast datasets, identifying patterns and relationships to create a personalized user experience. By delving into user data and preferences, these algorithms aim to deliver content tailored to individual interests, addressing the challenge of content discovery in an oversaturated digital landscape.

However, Gen AI tools and algorithms allow users to explore a broader spectrum of content, exposing them to diverse perspectives and nurturing creativity. Consider a scenario where a Large Language Model (LLM) powered chatbot is asked to provide three headlines for a specific article — repeating the same query yields a distinct set of answers. This variability underscores the potential of Gen AI in expanding user choice and facilitating access to diverse content and viewpoints. Moreover, it plays a crucial role in challenging the dominance of major social media platforms, loosening their control over the content we encounter online. The envisioned outcome is a future media landscape that is more competitive and democratized, fostering a richer and more dynamic experience for users than the current paradigm.

The Challenges of Generative AI for the Media Space

Content creators and companies are mastering Gen AI to augment and amplify their work. However, broader implications present challenges, such as generating content from licensed material, replicating voices and faces, and the potential for deceptive outputs. Balancing legitimate use and avoiding misuse introduces complexities requiring robust technical, legal, and regulatory frameworks. Navigating this complex landscape demands a nuanced understanding of the tools’ potential and vigilance against unintended consequences.

Concerns Around Data and Privacy

Media and entertainment companies heavily depend on personal and sensitive data to tailor content recommendations, offer targeted advertising, create interactive content, and develop innovative products and services. This dataset often includes personal details like names, addresses, financial information, and even sensitive particulars like medical records and social security numbers.

The intricacies of collecting and processing such data give rise to legitimate concerns regarding usage and accessibility. Companies employ algorithms to scrutinize consumer viewing habits and preferences, enabling targeted decisions on content delivery. Consequently, these entities must be acutely attuned to data privacy regulations governing personal information. In the realm of AI-driven data collection, adherence to GDPR standards is paramount. AI algorithms should be meticulously developed to minimize the collection and processing of personal data, prioritizing data security and confidentiality to navigate the intricate landscape of privacy regulations.

Copyright Issues

In a recent announcement, YouTube and UMG unveiled a pioneering partnership to establish a music AI incubator, prompting speculation about its intentions, with The Verge suggesting a focus on monetizing AI-generated music. YouTube CEO Neal Mohan countered this narrative by outlining the platform’s AI music principles, envisioning a framework that enhances creative expression and safeguards artists.

Committed to responsible AI adoption, the collaboration seeks to empower creativity while addressing copyright and AI-generated content challenges. As discussions on copyright law’s application to AI-derived music unfold, questions arise about the U.S. Copyright Office’s current stance, which requires a human author for copyright protection. The evolving landscape prompts considerations on the balance between AI and human creative contributions, shaping the future of copyright law.

Authenticity and Attribution in the Face of Synthetic Content Creation

Over the last two years, advances in Gen AI, particularly generative deep learning, have ushered in the era of synthetic media, thanks to platforms like MidJourney, RunwayML, Speechify, and deepfake tools like Reface. This accessibility, however, brings forth societal challenges, including diminishing trust in digital content, potential misuse by malicious actors, legislative limitations, and the growing need for authenticity.

As barriers to creating realistic synthetic media dissolve, especially in portraying well-known figures and places in fictional scenarios, our confidence in digital content faces erosion. Deepfakes blur the lines between reality and fiction, complicating the authentication of news and social media content.

While synthetic media unlocks possibilities in filmmaking, accessibility, art forms, personalized engagement, and R&D acceleration, it also raises concerns. Deepfakes, in the wrong hands, can fabricate events and manipulate statements, spreading misinformation rapidly through social networks and challenging the distinction between truth and fiction in public discourse.

AI-driven synthetic media offers great potential for progress, but without proper safeguards, it risks enabling harm. Collective action across technology, industry, government, and civil society is essential to maximize benefits and minimize dangers in this evolving landscape.

Ability to Create “Garbage” Content at a Huge Scale Using Gen AI

Edward Ginis expressed his greatest fear is the use of Gen AI to flood the internet with mundane, mediocre content, lacking any life or character, that offers no value to companies or consumers: “My biggest fear is noise and garbage at scales we have yet to see in the space. We will be flooded with the most automated, mundane music, news articles, and everything else. You’re going to start seeing a huge amount of generative AI content that will feel like everything else: mundane and flat without any new life or character, which often plagues AI versus what human capital can bring”.

For instance, Neil Clarke, the publisher and editor-in-chief of Clarkesworld Magazine, a Hugo Award-winning science fiction publication, recently halted submissions due to an overwhelming surge in artificial intelligence-generated manuscripts. Clarke explained the decision on Twitter, citing the inundation of AI-generated content.

The ease with which a single individual can employ Gen AI to churn out thousands of essays in one online session with minimal effort poses a potential problem. Scaled across millions of internet users could result in a deluge of generated content. It’s crucial to approach Gen AI responses skeptically, especially when dealing with seemingly factual information. Verifying details becomes essential, and maintaining a healthy dose of skepticism is your best defense when navigating generative AI content.

How to Approach and Invest in the Power of Generative AI

In March, KPMG rolled out its 2023 Generative AI Survey to cut through the hype and uncover practical insights into how enterprises can genuinely leverage Gen AI. The standout discovery: 77% of executives across diverse industries perceive Gen AI as the most impactful emerging technology, with 71% planning to implement their first Gen AI solution within the next two years.

While the potential applications of Gen AI in content creation, user engagement, software development, and data analysis seem boundless, the journey from hype to tangible business value still needs to be completed. As Gen AI evolves rapidly in its early stages, executives grapple with uncertainties surrounding security, reliability, job implications, and potential value before committing to widespread adoption.

IT managers must recalibrate expectations and strategies to optimize their talent pool effectively. Here are some key insights to navigate this landscape.

1. Unlock a deeper comprehension of generative AI opportunities by collaborating with advisors and tech partners. Given the novelty and potency of Gen AI, there’s a pressing need for thought leadership and ideation to enlighten and showcase its potential to media industry stakeholders. As Edward Ginis emphasized in a recent round table discussion, “The road to transformation requires vision, education, and innovation.” To maximize the impact, delve into your business’s most pertinent and valuable use cases, employing a validation framework for prioritization. Avoid isolating Gen AI; collaborate with experts familiar with your business domain and existing technologies. This collaborative approach ensures that Gen AI is seamlessly integrated, maximizing ROI.
2. Adopting a Solution Design-centric approach to building platforms initiates a crucial “discovery phase.” This phase extends the vision, delving into the risks associated with the products or solutions media companies aim to develop. It involves thoroughly exploring specific requirements, workflows, and a holistic understanding of the business objectives. When incorporating Gen AI technology, it becomes imperative to meticulously identify and measure risks and devise effective mitigation strategies before implementing the solution. This approach ensures a risk-aware foundation for successful platform development.
3. Embrace a proactive strategy for upskilling and reskilling your team to harness the potential of generative AI. Foster a culture of continuous learning by offering access to training programs emphasizing AI technologies, data science, and advanced programming. This approach ensures that your existing team members are well-equipped to leverage the benefits of AI effectively.
4. Cultivate an innovation-centric culture that champions experimentation and views failure as a vital step toward success. Empower team members to explore generative AI applications in their projects and recognize and reward innovative solutions. This environment ensures a dynamic and forward-thinking approach within your team

Conclusion

Generative AI is reshaping the media industry, accelerating automation and unleashing creative potential for artists. Collaborative efforts between significant platforms and industry players are decisive for responsibly addressing privacy and copyright concerns. Recognizing the impact of generative AI, executives must strategically recalibrate and upskill teams to bridge the gap from hype to tangible business value. In navigating challenges, individuals with a blend of technical and artistic skills emerge as key players, steering the future of the media and entertainment industry. Embrace this new era by fostering collaboration, skill development, and innovation in the intersection of technology and art. The future is in the hands of those ready to shape it.

Author: Doron Fagelson,
Vice President of Media and Entertainment Practice at DataArt
Originally published on https://www.dataart.com/blog/