Large action models are likely going to be a hot topic in the next year or couple of years.
Rabbit recently made an incredibly inspiring presentation, showcasing their large action model — a model that unlike large language models — doesn’t provide content but acts like an agent that can do tasks for you.
The way Rabbit describes it is very interesting and is the first time since Alan Kay from Xerox Park where there has been a complete change in the user interface of the devices we work with.
Rabbit states that our smart phones are plagued with tons of applications that over time have made user interface principles like navigation and way finding much more difficult.
Simple tasks have been decentralized into a mixture of expert applications in our phones. But easily access and executing tasks gets lost in a sea of icons.
Large action models could become the future operating system of most smart devices.
Where, instead of the user trying to understand how to do the “How” of actions, they simply have to tell the device “what” they want .
