Google recently shared insights into its Genie 3 generative AI technology at GDC, explaining the inherent memory limitations that cause its generated “world models” to begin displaying inconsistencies after approximately one minute of operation. This development marks a notable advancement, as earlier iterations of the world model were only able to maintain coherence for a mere few seconds.
Google’s Genie 3 AI World Models Encounter Limitations After About a Minute
By Gareth Wickshire
A Manchester-based gaming journalist with over eight years of experience covering the UK gaming scene. Started as a freelance writer for indie gaming blogs before establishing himself as a trusted voice in the industry. Specializes in breaking news about British game developers and emerging gaming trends. Known for his in-depth coverage of gaming events across England and insightful interviews with industry professionals.