Industry executives and experts share their predictions for 2025. Read them in this 17th annual VMblog.com series exclusive.
By Dominik Kovacs, CEO at Colossyan and Shahzaib Aslam, Director of Research at Colossyan
The past few years have been full of breakthroughs in artificial intelligence (AI). 2025 will follow this trend with the caveat that use and implementations in aspects like AI video creation will see an exponential growth across industries like learning and development (L&D), marketing and sales.
The new year promises to be a pivotal one, marked by increasingly sophisticated and interactive AI-driven video experiences. From advancements in conversational avatars and agentic AI to the fall of AR/VR training, 2025 will push the boundaries of innovation.
The rise of conversational and agentic AI avatars
Conversational avatars will be a significant breakthrough for the AI industry in 2025. They will not only be the next generation of AI chatbots, but they’ll be the next generation of interactive experiences in general, allowing for far more immersive experiences in a range of industries. However, widespread adoption relies on strong success metrics around the effectiveness of this tech and building trust around security.
These avatars are becoming highly feasible due to advancements in large language models (LLMs) and speech synthesis. Real-time, conversational AI, which includes both active listening and contextual understanding, is seeing applications in customer service, training, and content generation. This trend will continue to grow, particularly as models become more efficient and specialized, allowing for high fidelity interactions that feel natural.
The demand for hyper-realistic avatars will drive advances in generative AI, particularly in high-fidelity rendering and photo-realistic character generation. Developments in neural rendering, motion capture, and diffusion models are central to improving visual realism.
For mainstream adoption, user experience needs to be intuitive, secure, and adaptive to different devices. Improvements in computational efficiency and model compression along with improvements in the avatar fidelity will allow avatars to operate seamlessly across devices, addressing a key barrier to adoption.
Conversational avatars can be applied to various industries. In hospitality, guest relations trainees will be able to prepare for customer interactions with a realistic and responsive avatar on-demand. This will allow them to properly prepare for the dynamic nuances of real-world customer interactions, which can be difficult to replicate in a typical practice environment. The same goes for leadership training. A new manager who may struggle with sharing difficult feedback can practice with a conversational avatar to practice their delivery and understand how to deal with different responses, for instance. The use cases are endless.
Additionally, the concept of agentic AI avatars-avatars with autonomous decision-making abilities-is poised to gain traction. For avatars that leverage agentic AI to function effectively, advancements in reinforcement learning, memory, and long-context management are necessary. Early steps might involve avatars that can make basic decisions within specific contexts, like customer support, and evolve to take more complex actions over time.
The decline of AR/VR training
The AI video creation industry is fairly new, and so far, it has been focused on one-way communication via video. In 2024, videos were made interactive, but in 2025 these interactive experiences will reach new heights with the ability to converse with an avatar in real time.
However, AR and VR will not be integrated into interactive training. The momentum around AR/VR has been decreasing due to poor user experience and the demands of these technologies, and this will continue into the new year. The use of this tech also requires hardware, which is a significant barrier to widespread adoption.
Regulations and responsible AI development
Regulations around AI, particularly concerning high-risk systems and transparency obligations, are a necessary and welcome step toward ensuring responsible AI development. For the AI-empowered video industry, these evolving regulations will likely have a significant impact in two key areas in 2025: ethical AI implementation and user trust.
Regulations emphasizing transparency and accountability will drive companies to adopt practices like disclosing AI usage and ensuring generative models meet ethical standards, fostering trust and credibility among users and stakeholders. For AI-powered video platforms, this includes making avatars distinguishable from real individuals and maintaining audit trails for AI-generated content. While these regulations may initially slow adoption, particularly in high-risk applications such as regulated education or sensitive communications, they ultimately promote safer, more reliable AI deployment and create a competitive advantage in industries where authenticity is paramount.
Ultimately, 2025 will be a defining year for AI video creation and avatar usage, marked by both exciting advancements and critical challenges to overcome.
##
ABOUT THE AUTHOR
Dominik Kovacs is the CEO and founder of Colossyan, a leading text-to-speech AI video platform. Dominik founded Colossyan in 2020 to make video content more accessible for L&D teams, and the tool is now used by industry leaders across the globe, including P&G, BMW, Paramount, and more.
++
Shahzaib Aslam is the Director of Research at Colossyan, the AI video platform for workplace learning. Colossyan is a deep tech company with a separate research department dedicated to pushing the envelope in AI. Shahzaib leads the research team’s focus on human performance capture and generation, human behavior understanding, and high-fidelity rendering to create lifelike AI avatars that can be used to create engaging and interactive learning experiences.
