Increasing number of novel AI systems are being developed in a variety of domains including transportation, entertainment, conversational agents, etc. and are being introduced in the real world. For scenarios and edge cases that are difficult to bring about in the real world or pose an inherent risk, AI systems can benefit from evaluation in virtual and extended reality worlds (XR worlds) to demonstrate capabilities, evaluate performance, and anticipate shortcomings. Simulated environments are already an integral part of the testing framework in the development of autonomous vehicles. However, there is an opportunity for inventing new capabilities for the technology that produces the XR worlds, improving the quality of XR worlds, standardizing and democratizing the XR world creation process, and inventing novel techniques to utilize the XR worlds for training and testing AI systems.
The progress in XR technologies has led to the emergence of new forms of immersive remote social interactions beyond geographical boundaries. For the development of virtually embodied interactive AI systems with social intelligence within this context, it is crucial to evaluate such systems within interactive contexts with human participants. Here, XR worlds that augment the physical environment of the human participants hold promise.
Additionally, XR worlds so far have only been used in training humans; examples include the use of virtual audiences for coaching public speaking or applications in mental healthcare where virtual environments constitute a medium for self-exploration. With the rising presence of AI agents, training them using XR worlds also presents exciting challenges and opportunities. XR worlds provide a controllable, cost-effective, and safe environment for training complex AI systems and as such, this research space merits dedicated exploration.
In this Research Topic, we would like to explore the problem of using XR worlds to improve the quality of AI systems. This can be done in multiple avenues including but not limited to creating better virtual worlds with lesser resources and using the virtual worlds to improve AI systems.
We are interested in papers that:
- Explore new capabilities for creating XR worlds
- Develop new ways to improve and evaluate the quality and fidelity of XR worlds
- Standardize and democratize the creation of XR worlds
- Demonstrate the utility of XR worlds as a testbed for AI systems
- Develop new ways to utilize XR worlds for training AI systems
Keywords:
Virtual Reality, Artificial Intelligence, Machine Learning, Synthetic Data, Simulation
Important Note:
All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.
Increasing number of novel AI systems are being developed in a variety of domains including transportation, entertainment, conversational agents, etc. and are being introduced in the real world. For scenarios and edge cases that are difficult to bring about in the real world or pose an inherent risk, AI systems can benefit from evaluation in virtual and extended reality worlds (XR worlds) to demonstrate capabilities, evaluate performance, and anticipate shortcomings. Simulated environments are already an integral part of the testing framework in the development of autonomous vehicles. However, there is an opportunity for inventing new capabilities for the technology that produces the XR worlds, improving the quality of XR worlds, standardizing and democratizing the XR world creation process, and inventing novel techniques to utilize the XR worlds for training and testing AI systems.
The progress in XR technologies has led to the emergence of new forms of immersive remote social interactions beyond geographical boundaries. For the development of virtually embodied interactive AI systems with social intelligence within this context, it is crucial to evaluate such systems within interactive contexts with human participants. Here, XR worlds that augment the physical environment of the human participants hold promise.
Additionally, XR worlds so far have only been used in training humans; examples include the use of virtual audiences for coaching public speaking or applications in mental healthcare where virtual environments constitute a medium for self-exploration. With the rising presence of AI agents, training them using XR worlds also presents exciting challenges and opportunities. XR worlds provide a controllable, cost-effective, and safe environment for training complex AI systems and as such, this research space merits dedicated exploration.
In this Research Topic, we would like to explore the problem of using XR worlds to improve the quality of AI systems. This can be done in multiple avenues including but not limited to creating better virtual worlds with lesser resources and using the virtual worlds to improve AI systems.
We are interested in papers that:
- Explore new capabilities for creating XR worlds
- Develop new ways to improve and evaluate the quality and fidelity of XR worlds
- Standardize and democratize the creation of XR worlds
- Demonstrate the utility of XR worlds as a testbed for AI systems
- Develop new ways to utilize XR worlds for training AI systems
Keywords:
Virtual Reality, Artificial Intelligence, Machine Learning, Synthetic Data, Simulation
Important Note:
All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.