This week marked the Consumer Electronics Show (CES) in Las Vegas, a pivotal event where manufacturers and developers gather to unveil and introduce their products globally. This year, the excitement centered around artificial intelligence, with the Rabbit R1 emerging as the product that was the buzziest.
The large action model (LAM), is a new type of foundation model that Rabbit claims is able to understand and execute human intentions on computers. It is fundamentally different from traditional AI assistants like Siri or Alexa in that it can not only understand what you are saying but also take actions to complete tasks on your behalf. This is made possible by LAMs ability to learn the interfaces of different apps and then trigger actions within those apps using natural language.
For example, if you ask Rabbit to book a flight, it will be able to open the Kayak app, search for flights that meet your criteria, and then book the flight for you, all without you needing to touch the screen. Unlike traditional AI assistants, which would typically only provide you with a list of possible flights or open the Kayak app for you.
What might be the most interesting feature, according to the video, Rabbit R1 learns by observing and recording a bunch of usage of different apps, and then it abstracts that away with a large language model. It can learn how to get to a specific function within an app, such as playing a SteelPulse or Bob Marley song on Spotify, by recording how a human user would do it.
Rabbit R1 shows that it is possible to build a device that can understand and execute human intentions on computers, and there this is a much better way to interact with computers than the current app-based system.
This is an amazing step in a totally new direction – look for responses from Apple, Samsung, Google and others to respond.
... and yes I ordered one.
Komentarze