＂AI moment＂ LAM big action model rises! This is the AI we want!
In the quiet, the landmark GPT-3.5 has been published for one year. During this year, we not only witnessed, but also personally participated in the historic changes in science and technology. Various AI dialogue applications based on Large Language Model (LLMs) have spread all over our lives. Many people no longer rely on opening search engines to solve problems, but turn to dialogue with AI for answers.
By Midjourney
However, many small partners feel a little bored after using it for a period of time. Indeed, the current AI can talk to you about Kan Kan in the dialog box, but it can’t help you solve some practical problems. For example, book a plane ticket for you, call a taxi, or when you want to remove the background of this photo in PS, these dialogs will only give you specific steps, but they can’t help you complete them. Many users are eager for a more hands-on AI, which can not only provide information, but also perform specific tasks.
By Midjourney
Laziness is the primary productivity of human beings. It is this desire for comfort and convenience that breeds a series of epoch-making products. Nowadays, in order to satisfy people’s desire that AI can help solve practical problems, large-scale action models (LAMs) came into being. Their goal is to make AI’s ability not only limited to providing information and performing virtual tasks, but also extended to physical actions in the real world, further releasing the bondage of human beings from daily chores.
By DALL-E 3
At the beginning of December 2023, an innovative research team, Rabbit, released their first large-scale action model system-Rabbit OS.
The picture comes from the Internet.
The Rabbit team explained that Rabbit OS is not built on a large-scale language model like ChatGPT, but is innovatively built on a "large-scale operation model". It is conceptually similar to the Google Assistant class. Rabbit OS can control music, order a takeaway, make a drip, buy a plane ticket, buy a movie ticket, buy a bunch of flowers for his girlfriend and buy daily necessities through a unified interface. Users don’t have to switch between different applications or log in repeatedly-just tell Rabbit OS what you need and it will help you.
The picture comes from the Internet.
The picture comes from the Internet.
After the concept of LAM was released, it attracted strong attention from foreigners. After all, this logical chain is completely correct. When LLMs is mature, people only need to talk to AI, and AI can understand your intention accurately and clearly, and directly realize the practical operation you want through LAMs.
The picture comes from the Internet.
The picture comes from the Internet.
The picture comes from the Internet.
So Rabbit released the first Rabbit R1 loaded with LAMs Rabbit OS.
The picture comes from the Internet.
Rabbit R1 looks a bit like a miniature version of the knob TV in the 1990s. Its rounded appearance is the result of deep cooperation with Teenage Engineering, a well-known design company. Two-thirds of the front left side is the screen, and the other part is the control area and camera, which looks very retro.
The picture comes from the Internet.
The compact R1, which is only half the size of the iPhone, is equipped with a 2.88-inch touch screen and a rotating camera, making it easy to take photos and videos. In addition, through the unique wheel/button, users can easily browse the content or interact with the built-in assistant of the device.
The picture comes from the Internet.
The picture comes from the Internet.
In terms of performance, Rabbit R1 has a 2.3GHz MediaTek eight nuclear processor and uses a storage combination of 4GB+128GB, which is enough for this small device.
The picture comes from the Internet.
The price of this equipment is 199 dollars (about 1450 yuan), which is very conscience for the first generation products of a new company. The 10,000 units of the first generation mass production were also sold out in the afternoon after the release.
The picture comes from the Internet.
So, what can Rabbit R1 do for users? According to the introduction of foreigners who have already used it, Rabbit R1 is a voice interactive intelligent device. Users can start playing music software on their mobile phones or take taxis automatically without opening applications, which is a bit like the realization logic of walkie-talkie, as if there is a private secretary opposite the walkie-talkie really helping you realize all this.
The picture comes from the Internet.
Rabbit R1 connects the application with the user’s online account, and uses its self-developed large-scale action model to analyze voice commands, supplemented by GPT language model (mainly responsible for converting human commands into text and interpreting them) to complete the task. It allows users to customize personalized operations through voice commands without developing a specific API. The action model of the device learns the operation method by analyzing the interaction between people and applications.
The picture comes from the Internet.
Let AI move from dialog box to reality, which may be the significance of Rabbit developing LAM and releasing Rabbit R1. The materialized operation logic and unique action learning function have made Rabbit R1 continuously praised overseas.
The picture comes from the Internet.
The picture comes from the Internet.
In this era of AI generate, major AI technology companies have launched their own large-scale language model (LLM) applications, such as GPT, Bard, and domestic classical Chinese, such as One Heart, Dark Side of the Moon, and Thousand Questions on Meanings. The diversity of these products often confuses users. Rabbit’s strategy is completely different. It connects different applications through an interface to solve users’ practical problems. It is similar to a multifunctional super application. Users can complete almost all tasks through this platform and experience one-stop convenient AI services.
The picture comes from the Internet.
The picture comes from the Internet.
Looking forward to the future, we expect to see the integration scope of LAMs (Large Action Model) extended to more smart home devices, smart cars and smart public facilities. This integration is expected to realize a real man-machine conversation-the device can automatically recognize, understand and perform the required operations simply by the user issuing verbal instructions. This will be a huge leap forward, which will greatly improve the convenience and efficiency of life, and will also provide greater autonomy and support for the disabled and other groups that need assistive technology.