Ethical Leadership in AI: How Taehyung Yoo’s Melting Pot AI is Redefining the AI-Artist Relationship
In today’s digital age, the intersection of technology and ethics continues to draw attention. I recently had the pleasure of interviewing Taehyung Yoo, the CEO of Mirrorboard, who is also at the helm of the innovative startup, Melting Pot AI. Yoo shared insights into his unique journey, from social campaigner to tech entrepreneur, and his vision for a symbiotic relationship between AI and the arts.
Can you introduce yourself and tell us about your background before founding Melting Pot AI?
I am Taehyung Yoo, the CEO of Mirrorboard, currently operating Melting Pot AI. Before starting my own company, I was a campaigner interested in various social issues and led numerous campaigns. To address the low birth rate in South Korea, I organized a massive speed dating flash mob called ‘Solo Conquest’ with the participation of ten thousand singles.
Additionally, to highlight issues with wages and employment, I initiated a project called ‘Yoo Taehyung for Sale,’ where various companies bid on my annual salary, creating an issue by rejecting the highest bid. In the ‘Do You Know Obama?’ campaign, I managed to deliver a letter to President Obama through a friend of a friend, suggesting the possibility that everyone can be connected to the highest powers.
These experiences motivated me to start my own startup. I wanted to demonstrate that businesses can grow by consumers choosing products with the right values. Melting Pot AI is a natural extension of the social campaigns I have previously conducted.
What inspired you to start Melting Pot AI, and what stands at the core of your business philosophy?
Melting Pot AI is a startup founded with the goal of solving all inefficiencies and problems in the world. Our core vision is the ‘coexistence of artists and AI,’ a concept naturally formed during the process of identifying opportunities in the market. The unique reason we developed the generative AI platform, Melting Pot AI, is because we place greater importance on ‘conscience’ and ‘courage’ than others.
Initially, we quickly entered the market using generative AI models. Our first product was a prompt marketplace, and our second was a stock image creation service that produces images resembling real photos. This stock image service became popular, especially among marketers and creators looking for royalty-free images.
However, after achieving success, we began to deeply consider the impact of AI replacing the roles of traditional photographers, models, and studios in their lives. During this process, we learned about the decline in revenues for stock image creation communities and platforms like Shutterstock and Getty Images. It was particularly shocking to learn that artists at Shutterstock were receiving a one-time payment of $0.01 per image.
Further discussions revealed that designers, composers, and writers working for gaming and content companies were being laid off en masse. This situation highlighted the harsh reality of AI replacing human jobs, causing us to feel significant guilt.
As we delved deeper into this issue, we questioned whether generative AI was evolution or robbery. Our team extensively researched the social and cultural impacts of AI, discovering not only that AI threatens the jobs of artists, but also that it faces significant sustainability issues, such as the ‘model collapse phenomenon.’
Ultimately, we concluded that it is necessary to create an ecosystem where AI and artists can protect and develop together. To this end, we are moving towards a system that respects artists’ works and ensures appropriate compensation, enabling AI technology to enhance the creativity of artists. Melting Pot AI pursues this vision, aiming to create greater cultural value by fostering a sustainable way for AI and artists to develop together.
How does the opt-out mechanism function within Melting Pot AI, and what are the technical and ethical challenges associated with implementing such a system?
Melting Pot AI employs a robust system to rapidly identify matches between the artworks submitted by artists and the datasets used by other generative AI developers. We actively monitor these AI developers and conduct periodic verification requests to companies included in our monthly list. This monitoring is crucial to determine whether the data of Melting Pot AI members has been used without authorization and to actively demand that such data be opted out from the AI training datasets.
Currently, there is no mandatory requirement for AI companies to disclose their training data. However, with the impending implementation of the EU’s AI Act, which will likely mandate greater transparency in the use of training data, our monitoring activities are set to become more vigorous.
Once the AI Act is fully operational, it will enable more proactive enforcement of data usage standards, ensuring that the rights of artists are protected and that their contributions are not used without proper authorization and compensation. This proactive approach not only helps maintain the integrity of artists’ work but also sets a precedent in the industry for responsible and ethical AI development practices.
In what ways has the recently published EU AI Act influenced your approach to building an ethical AI platform, and how does Melting Pot AI go beyond basic compliance?
Melting Pot AI is a project that was prepared in anticipation of serious AI and copyright issues before the EU announced its AI Act. The significance of our project lies in creating a platform system that inherently ensures that the generative AI used is legal, rather than merely ‘proving’ its legality.
The AI Act briefly addresses copyright issues, stating that data must be collected in compliance with copyright law. Many AI companies will refer to ‘fair use’ in this context. Fair use allows the use of copyrighted data without the consent of the copyright owner, but it must serve a public purpose and not infringe on the rights of the copyright holder. The Napster case is a notable reference for this.
Over the past year, there have been lawsuits filed by artists against generative AI companies. Not once have artists won these cases, and generative AI companies have been adamant in asserting their legality based on this principle.
Melting Pot AI is completely free from these problems. Artists directly participate in our platform, give their consent, and provide the data for learning themselves, thus eliminating the need for us to prove fair use.
How do artists typically engage with the platform, and what strategies have been effective in encouraging their participation and trust in the system?
Artists can easily engage with Melting Pot AI by uploading the works they want to provide as training data. After the upload, the AI training and subsequent royalty settlements are automatically processed. In simple terms, artists just need to upload their work, wait, and then receive their compensation. Additionally, artists who upload their artwork can access the website’s forum and connect with other artists worldwide and with Melting Pot AI’s HQ directly.
To foster trust among artists, we offer an additional service: an AI copyright management service. This service periodically monitors whether the works submitted to Melting Pot AI are being used for training by other AIs. If any unauthorized use is detected, the service will handle opt-out requests and claim compensation on behalf of the artist.
Once the AI Act is enforced, AI companies will be obligated to disclose their training data. We plan to gather the artworks from our artists and officially inquire about potential copyright violations to other generative AI companies. This proactive approach not only protects our artists but also reinforces the integrity and transparency of our platform.
How does Melting Pot AI respond to the challenges posed by other platforms facing legal scrutiny for uncompensated use of creative works?
At Melting Pot AI, we provide royalties to artists based on the ‘actual use’ of their data. For example, if a photo of a deer is generated through Melting Pot AI, royalties are paid to the artist who provided the original deer photo used in training the AI.
The settlement method plans to adjust up to 80% of the AI’s net profits to the artists. This generous royalty model not only ensures that artists are fairly compensated for their contributions but also aligns with our vision of supporting and fostering a sustainable ecosystem where artists and AI can coexist and thrive. This approach demonstrates our commitment to ethical AI practices and respect for artistic contributions.
What specific measures are you taking to ensure transparency, accountability, and fairness in AI governance across Melting Pot AI’s operations?
Melting Pot AI takes thorough measures to ensure that only genuine artists who actively engage in creative activities are admitted as members. Additionally, we employ an internal engine to monitor the data pool collected through artists, ensuring that there are no cases of copyright infringement or duplicate data.
One of our most valued principles is ‘fairness.’ We believe that simply calculating royalties based on the rate of data used in training is not equitable. This is because some artists may provide popular images, others might provide very rare images, and some might provide common and ordinary images.
To address this, Melting Pot AI employs a technology that tracks the raw data used every time an image is generated to accurately assign royalties. This feature is unique to Melting Pot AI and uses our proprietary technology called 2layered labeling, which significantly reduces the computing power needed compared to traditional GDI (Generic Data Inversion) technologies for tracing raw data. This system ensures a more equitable distribution of royalties, recognizing the unique contribution of each artist’s work.
Are there specific partnerships or collaborations that Melting Pot AI is pursuing to broaden the impact of ethical AI practices in the industry?
Melting Pot AI is not only focused on developing its own generative AI but also aims to instill a sense of copyright sensitivity throughout the entire industry. We believe that any company interested in adopting Melting Pot AI’s ethical generative AI development methodologies should be able to collaborate with us in business.
This open approach helps spread ethical practices and standards across the AI industry, encouraging other companies to prioritize artist rights and fair compensation. By sharing our methodologies and technologies, we aim to create a more responsible AI landscape where innovation coexists with respect for intellectual property and artistic contributions.
How do you see ethical practices influencing the evolution of generative AI technologies, especially in the context of artistic creativity?
As the market for generative AI grows, various technical issues arise, including the phenomenon known as “model collapse.” A significant problem is the decreasing availability of data for AI to learn from, which is primarily created by artists.
From the perspective of AI development, protecting artists is not merely optional—it’s essential for the continuous advancement of AI technology. By ensuring that artists are incentivized to keep producing high-quality, original content, AI systems can have access to diverse and robust data pools that enhance their learning and functionality.
This symbiotic relationship highlights the importance of establishing and maintaining ethical practices within the AI industry. It stresses the need for a framework where both technology developers and creative individuals benefit mutually, ensuring the sustainability of both AI innovation and artistic creation. This approach not only supports technological advancement but also upholds the value and rights of the artists contributing to these advancements.
What exciting or innovative applications of Melting Pot AI’s ethical platform can we expect to see shortly?
For effective model training in AI, a substantial amount of data is required. Melting Pot AI plans to begin its first model training only when the collection reaches 100 million images submitted by artists. This massive dataset is essential to ensure that the AI can learn from a broad and diverse range of inputs, which enhances the model’s ability to generate high-quality and varied outputs.
To achieve this goal, we encourage a large number of artists to participate and contribute their creative works. Each submitted image not only enriches the training set but also plays a crucial role in advancing the field of generative AI. By contributing, artists are not only supporting the development of cutting-edge technology but are also ensuring that their creative rights and contributions are respected and compensated within this growing technological landscape.
Lastly, how do you see Melting Pot AI influencing the broader discussion on ethical AI and shaping the industry?
Generative AI, like that developed by Melting Pot AI, represents a paradigm shift in how people engage in creative activities, much like the impact of the internet and mobile technology. As technology profoundly affects many people, there is an increased need for fairness and ethical consideration in its development and deployment.
Melting Pot AI aims to set a precedent by establishing a point of connection between artists and AI, ensuring that as generative AI continues to evolve, it does so in a way that supports and includes the artistic community. This approach not only aids in the technological advancement of AI but also protects and promotes the creative rights of the artists whose work fuels these developments.
By fostering a balanced ecosystem where artists are respected and compensated fairly, Melting Pot AI champions a model where innovation in AI goes hand in hand with ethical practices, setting a standard for the industry to follow. This ensures that as generative AI transforms creative processes, it does so while enhancing and not exploiting the contributions of human creativity.
Yoo’s vision for Melting Pot AI encapsulates a future where technology and human creativity coexist in harmony, enhancing each other’s capabilities and ensuring that digital evolution remains inclusive and equitable. As the AI landscape continues to evolve, it is clear that ethical considerations will play a pivotal role in shaping its trajectory, with Melting Pot AI at the forefront of this transformation.