Earlier this month, I experienced a remarkable digital breakthrough. During a press briefing preceding CES, Nvidia showcased a demonstration of its Ace microservice, an AI suite with the ability to create fully voiced AI characters. I was captivated as a presenter interacted with an in-game NPC using a microphone, only to receive real-time responses from the digital character. It was undeniably a feat straight out of science fiction, prompting one burning question: How did this technology acquire such capabilities?
Nvidia’s response to this query was somewhat cryptic, stating that there was “no simple answer” to the question. This vague statement triggered a wave of speculation on social media, with users assuming the worst-case scenario. Rumors began circulating that Ace might have been trained on unauthorized content owned by Nvidia. Although Nvidia later clarified that they only utilized approved data, the lingering ethical and artistic dilemmas left gamers understandably wary.
Amidst the onlookers observing this unfolding drama was Purnendu Mukherjee, a software engineer with a pivotal role in the AI technology at the center of this controversy. Mukherjee, the founder of Convai, the generative AI company powering Nvidia Ace, took it upon himself to address the misconceptions surrounding his creation rather than remain a passive bystander.
In a candid discussion with Digital Trends, Mukherjee delved into the ethical implications of AI tools like his creation. From concerns about job displacement to fears of AI diluting the essence of art, Mukherjee shared his optimistic vision of a future where AI collaborates harmoniously with artists to bring their creative visions to fruition. However, when delving into the contentious issue of data usage, his explanations seemed to raise more questions than they answered.
Can AI and artists collaborate effectively?
Mukherjee’s journey into the realm of AI was fueled by a childhood curiosity about the human mind and a later fascination with deep learning during his studies in India. After a stint at Nvidia and pursuing further education in the U.S., Mukherjee ventured out on his own to establish Convai in April 2022, self-funding the venture for ten months.
A passionate gamer from his early days, Mukherjee envisioned the potential of AI to enhance gaming experiences, drawing from his experiences playing competitive titles like Counter-Strike. This vision materialized into a groundbreaking innovation that leverages various AI processes to create fully voiced NPCs capable of dynamic interactions with players in real-time. His objective? To elevate the level of engagement in games.
Mukherjee pointed out how renowned games like Baldur’s Gate 3 and The Witcher feature captivating narratives that players often only scratch the surface of due to limitations in NPC interactions. With current technology, these NPCs could evolve to engage players on a deeper level, providing additional insights while maintaining character authenticity—an aspect Mukherjee is keen on exploring.
Throughout our conversation, Mukherjee reiterated his belief that generative AI tools would not diminish the need for artists but rather amplify their roles. He emphasized that AI tools would necessitate extensive input from artists to ensure optimal training of the technology. This symbiotic relationship, according to Mukherjee, would lead to a surge in demand for narrative designers and potentially higher compensation for voice actors involved in training these AI tools.
Mukherjee’s optimistic outlook on the impact of AI on the gaming industry, despite the ongoing wave of layoffs, underscores his conviction that adaptation and collaboration with AI are essential for artists to thrive in this evolving landscape.
The nuances of data utilization
The discussion steered towards the pivotal role of data in training AI models, a topic that often sparks controversy and legal disputes. Mukherjee advocated for fair compensation for contributors to data sets used in AI training, emphasizing the ethical sourcing of data, particularly for commercial applications.
Addressing concerns about Convai’s data practices, Mukherjee affirmed the company’s adherence to utilizing only authorized data sets. He highlighted the meticulous process of vetting base models from reputable sources like OpenAI, ensuring compliance with commercial licensing and ethical standards. Mukherjee underscored the collaborative efforts with voice actors in text-to-speech applications, emphasizing the increased demand for voice actors in their AI training endeavors.
However, the mention of Convai’s reliance on models trained by OpenAI raised questions about potential copyright infringement issues. Mukherjee clarified that while Convai utilizes models from OpenAI, they do not directly access the underlying data. This distinction, though subtle, forms the basis of Convai’s stance on ethical data usage.
As the conversation delved deeper into the intricacies of data dependencies among AI models, Mukherjee acknowledged the complexity of tracking data sources across different models. The interwoven nature of AI development, with tools built upon existing frameworks like OpenAI, poses challenges in discerning the origins of specific data sets—a conundrum that Mukherjee candidly acknowledged.
Embracing regulatory frameworks
In light of the evolving AI landscape, the topic of regulatory oversight emerged, prompting a discussion on the need for balanced regulations to guide technological advancements without stifling innovation. Mukherjee recognized the necessity of regulations to govern AI applications responsibly, drawing parallels to the regulation of automobiles to ensure public safety.
While acknowledging the transformative potential of AI, Mukherjee remained steadfast in his belief that the benefits of AI outweigh the risks, provided that ethical considerations remain paramount. He envisions a future where AI tools complement and empower artists, dispelling the notion of AI as a threat to creative professions.
As our conversation concluded, Mukherjee expressed gratitude for the opportunity to clarify Convai’s contributions amid the media coverage of Nvidia Ace. The conversation ended on a reflective note, highlighting the parallels between artists feeling overshadowed by AI tools and Mukherjee’s own quest for recognition in the AI landscape—a poignant realization underscoring the intricate dynamics between technology and creativity.
