General World Models Guidance Group no. 2: Exploring the Future of AI Driven Content Creation
On May 11, researchers, creative professionals, media companies, and technology innovators gathered at Ghent University for the second Guidance Group meeting of the General World Models (GWM) project. Together, they explored how emerging AI technologies and General World Models could transform audiovisual production, discussed key challenges and opportunities for the industry, and helped shape the future direction of the project through valuable feedback and insights.The session combined project updates, practical demonstrations, emerging research, and critical discussions about the future role of AI in content creation.
Building a Community Around General World Models
The day started with a project recap and an overview of the ambitions of the General World Models project. Running from October 2025 until September 2027, the project aims to help companies understand, evaluate, and adopt next generation AI technologies through research, experimentation, proof of concepts, and knowledge sharing. Participants were introduced to several new members joining the ecosystem, including Hotel Bonka/Hotel Talent, Photostudio.IO, and Postmen BV. The growing network highlights the increasing interest from industry partners in exploring how AI can transform audiovisual production workflows. The team also presented new resources such as: the ‘Creative AI Playbook’, a curated mapping of new AI-tools and presentations to be found under the section ‘Learning Page’ on this website!
From Research to Real World Applications
A key focus of the meeting was demonstrating how research is being translated into concrete proof of concepts.
Three major research tracks were presented:
- AI powered video production workflows
- AI generated 3D content
- AI driven embodied avatar interactions
These proof of concepts will evolve throughout the project and serve as practical testing grounds for industry applications.The audience also received an update on student projects exploring photorealistic, 2D, and 3D AI avatars. One example showcased the development of an intelligent virtual energy advisor, illustrating how generative AI, character creation, and conversational systems can be combined into user facing applications.
AI in Video Production: What Works and What Doesn’t?
One of the most anticipated sessions focused on the current state of AI in video production. Drawing on data from the Video Experience Day 2026 survey, researchers presented a clear trend: AI adoption within video production continues to accelerate. According to the survey, 87% of respondents now use AI somewhere within their production workflows, a substantial increase compared to previous years.
The discussion revealed that companies primarily use AI for three purposes:
- Improving efficiency and accelerating workflows
- Supporting concept development, ideation, and scripting
- Enhancing post production processes such as upscaling, audio cleanup, subtitling, and visual effects
At the same time, participants examined how the AI landscape itself is changing. While tools such as Midjourney, Adobe Firefly, ElevenLabs, and Runway dominated previous years, newer systems such as Google Veo 3, Seedance, Nano Banana, Kling, Gemini, and ChatGPT are rapidly becoming part of everyday production pipelines.
Comparing Three Production Approaches
A particularly valuable part of the session compared three different production workflows:
Traditional Production
Traditional workflows continue to offer the highest level of manual control and predictability. However, they remain labor intensive and require significant production resources.
Cloud Based AI Production
Cloud based AI tools significantly reduce production effort and team size requirements. Researchers demonstrated workflows combining Midjourney, Nano Banana, image generation systems, and AI animation tools. While highly efficient, these workflows often introduce waiting times and depend heavily on external model providers.
Real Time AI Production
The final workflow explored real time AI production using technologies such as custom LoRAs, Unreal Engine, ComfyUI, PAIC, and advanced AI pipelines. Real time workflows offer immediate creative feedback and enable decisions to be made during production itself, reducing post production effort. However, technical stability and image quality remain ongoing challenges.
The comparison highlighted an important conclusion: there is no single “best” workflow. Instead, creators increasingly combine traditional methods, cloud based AI, and real time AI depending on project requirements, budgets, and creative goals.
The Rise of AI Generated 3D Content
Another major theme of the day was the rapid evolution of AI generated 3D content.
Participants received an overview of the latest developments in AI assisted 3D creation, including emerging workflows for character generation, environment creation, and interactive experiences. The session emphasized how advances in 3D generation are increasingly converging with video generation and simulation technologies, creating new opportunities for immersive storytelling and digital production.
Why 3D Engines Matter for General World Models
The discussion then shifted toward one of the project’s core research questions: why 3D engines are becoming increasingly important in the development of General World Models.
Researchers argued that language models alone are insufficient for capturing the complexity of real world interactions. Instead, spatially aware systems and 3D environments offer a richer foundation for understanding movement, causality, interaction, and user experience. By combining AI with real time simulation environments, future systems may be able to learn and reason about the world in fundamentally different ways.
Adaptive and Personalized Experiences
The morning concluded with presentations exploring adaptive technologies in XR and education.
Researchers demonstrated how machine learning and AI can be used to personalize a VR emotion regulation application, dynamically adapting experiences to individual users. Another presentation focused on adaptive virtual tutors that can provide personalized feedback in real time, illustrating how AI can move beyond content generation toward more responsive and individualized interactions.
A Critical Conversation About Ownership and AI
After lunch, participants attended a seminar led by Julie Van Pée and Matthias Van Damme examining one of the most pressing questions facing the creative industries today:
Who owns content in an AI driven world?
The session explored the legal, ethical, and societal implications of AI generated content, addressing topics such as authorship, intellectual property, training data,
copyright, and the responsibilities of creators and organizations working with generative technologies. We also would like to refer to our allready existing white paper on the legal aspects made by Timelex lawyers that can be found in our Creative AI Playbook: Whitepaper – Legal Aspects
Looking Ahead
The second Guidance Group meeting demonstrated both the enormous potential and the practical challenges of AI adoption within the audiovisual sector. Rather than focusing on hype, the discussions emphasized experimentation, critical reflection, and real world applicability.
As General World Models continue to evolve, the project aims to provide companies with the knowledge, tools, and practical insights needed to make informed decisions about the future of AI driven production.
With new proof of concepts underway, upcoming workshops, inspiration sessions, industry residencies, and the Video Experience Day 2026 event already on the horizon, the project is building a growing community dedicated to understanding and shaping the next generation of creative AI technologies.
See you soon!
