DroppGroup & UC Berkeley Collaborate to Pioneer AI Media Benchmark

Leading generative digital media developer droppGroup is thrilled to announce an extension of its current partnership with UC Berkeley, which is housed under the latter’s Launchpad organization.

Through the collaboration, a benchmarking framework for droppGroup’s cutting-edge generative media intelligence (GMI) system—which creates multimedia content created by artificial intelligence (AI)—including 3D models, virtual reality (VR) experiences, and metaverse environments—has been developed. The framework, which will be the first of its type, will make it possible to evaluate droppGroup’s GMI system consistently across a range of parameters, including user engagement, performance, safety, and efficiency.

The generative AI industry is expected to grow from $11.3 billion in 2023 to around $52 billion by 2028. Two prominent model types—generative media intelligence (GMI) and large language models (LLMs), such as those used by OpenAI’s ChatGPT—are responsible for the industry’s explosive growth. While text-based activities have been transformed by LLMs, GMI is raising the bar in media content production by providing artists the tools to realize their ideas in previously unthinkable ways.

GMI is now used by the droppPhygital content creation platform via droppaMiGO, its proprietary system. This system creates complex 3D models and immersive digital worlds from basic inputs like photos by using GMI.

Beyond static media, this development makes it possible to create dynamic and interactive digital assets that are immediately usable in virtual reality, gaming platforms, and digital markets. Moreover, aMiGO drastically cuts down on the time and knowledge needed to create this kind of content.

Nevertheless, this represents a revolutionary change from the labor-intensive, conventional digital content development procedures to an effective, AI-driven paradigm. The capacity to quickly create intricate, interactive digital assets has broad ramifications for e-commerce, gaming, entertainment, and education sectors, changing the way people engage with digital information. A notable gap exists in autonomous 3D content production and 3D environment modeling: there is currently no benchmark or standardized system. Because of this discrepancy, evaluating the effectiveness, precision, and efficiency of generative AI systems in producing intricate, interactive digital products is difficult.

As a result, droppGroup is working with Launchpad, a leading machine learning group at UC Berkeley, to create an industry-standard benchmark for 3D and 3D environment modeling that is automated. This project not only closes the gap in specifications but also establishes a standard for the rigorous assessment of generative AI capabilities. It serves as a cornerstone for future advancements and offers an organized means of tracking advancements and enhancements in generative AI technology.

Christopher Kelly, President & Co-Founder of droppGroup, discussed the cooperation as follows:

“This partnership exemplifies the powerful outcomes achievable when academic research and industry innovation converge. Combining UC Berkeley’s academic excellence and research capabilities with droppGroup’s cutting-edge technology and real-world applications, this collaboration is set to push the boundaries of what’s possible in AI and machine learning. It demonstrates a successful model for how such collaborations can accelerate technological advancements and applications in society.”

DroppGroup and UC Berkeley’s Launchpad are setting the foundation for a new age of AI-driven digital creativity by demonstrating how effectively AI systems can comprehend textual inputs and translate them into high-quality, interactive 3D models and settings. More generally, the project seeks to establish new benchmarks for AI technology development and assessment.

With a comprehensive timeframe spanning from February to the end deliverable in May 2024, the benchmarking project formally begins in the first quarter of 2024.

This innovative project is an extension of dropp and UC Berkeley’s continuous collaboration, which was started in 2022. Since then, the companies have worked to combine dropp’s cutting-edge and well-established ML/AI tech stack with the core strengths of the Launchpad team in machine learning and artificial intelligence.