In recent days, I believe that many friends have seen about itGPT Store, Vision Pro, Rabbit R1, AI Pin, NVIDIA Ace (**ATAR Cloud Engine), DingTalk Personal Assistant, Honor MagicOS 80and other AI agents that are deeply integrated with AI technology or platforms that carry AI agents. Some are related to personal applications, such as DingTalk Personal Assistant and Honor MagicOS 80 is the application for the individual; Some are related to corporate organizations, such as Nvidia Ace (**ATAR Cloud Engine) that can be used by game companies to improve the gaming experience.
Apple's Vision Pro can be used by individuals or companies as a further business application; GPT Store is an Apple Store-like platform store developed by OpenAI, which can provide an integrated app store for individual or enterprise developers, it is not an AI agent per se, but it is a platform that allows users to create and customize AI agents.
In the current technological environment, the field of application of artificial intelligence** (AI agent) is rapidly expanding to cover all aspects of our daily life and work. Behind this trend is the continuous innovation and efforts of major companies in product development and business competition.
Why are there so many AI agent releases at the beginning of 2024? Just as at the beginning of 2023, the intensive release of various large models, recalling that the intensive explosion of large models in 2033 is also due to the rapid development of big data and large computing power, and the outbreak of AI agent in early 2024 also stems from the maturity of software and hardware technical conditions, large-scale market demand, and rich and diverse data resources.
If the intensive explosion of large models in 2023 is the first half of AI, then the intensive release of AI agent in 2024 belongs to the second half of AI. Different from the first half, everyone is busy with the competition of the basic model and the application model, the second half is compared to the real market side, if the users in the first half are B-end customers, then the users in the second half are gradually transferred to the C-side, the C-end is facing the world's 5 billion social ** users, and the high-quality AI agent will be no less than the revolution of Apple's mobile phone to the mobile phone industry.
The C-end market is also a real rich mine, throughout the world's high-capitalization companies are all C-end markets without exception, and the users of the C-end market are also the most picky, because the C-end is oriented to different individuals, and it is really difficult to unify the needs and hobbies of different individuals, but because of this, high-quality product research and development companies will really be committed to the struggle for human development and develop products that everyone likes, and 2024 is destined to be the beginning of fierce competition for AI agents.
On January 10, 2023, Honor released the industry's first human-computer interaction (IUI) operating system based on AI intent recognition - MagicOS 80 (Magic OS 8.)0）。This new operating system integrates Honor's self-developed platform-level 7B device-side AI model, the Magic model, which not only strengthens the kernel of the operating system, but also provides comprehensive AI capability support for IUI.
The main difference between platform-level AI and application-level AI is that the former can be used as a technical foundation to fully empower the operating system and is regarded as a new kernel for the next generation of operating systems. The traditional operating system kernel is mainly responsible for managing and scheduling the hardware resources of the system, such as GPU and memory, to ensure the normal operation and efficient utilization of the system. However, with the growing demand from users, AI systems need to parse many human-related factors to achieve a truly human-centric experience.
Traditional operating systems cannot effectively calculate and process three types of human-related factors: personal knowledge base, human position and state perception, and human habits and portrait learning. Therefore, we need a completely new kernel to meet these needs. The power of platform-level AI lies in its ability to manage and handle multiple personal factors to help the operating system accurately identify user intent.
With this capability, the new operating system can bring everyone a smart experience of "guess what you think, know what you need", which is undoubtedly a big step forward in human-computer interaction. This also indicates that we are moving towards a new era, an era in which artificial intelligence is deeply integrated with human life.
Also remember the 2007 Apple mobile phone press conference, Jobs once showed the history of Apple's revolutionary user interface, from the computer mouse, to the iPod click wheel, to the first generation of iPhone multi-touch screen, the mobile phone industry revolution, if the Apple mobile phone as the representative of the multi-touch screen technology is the first innovation of human-computer interaction of the mobile phone, then to the glory magicos 80 as an example, AI large model technology is the definition of the second human-computer interaction of mobile phones.
MagicOS 8., which is redefined by the AI model0 is a comprehensive personal assistant, it can plan a day's itinerary for you, and make efficient arrangements for your life and work, whether it is a taxi ride, hotel accommodation, or a personal work arrangement plan, in short, it includes all aspects of food, clothing, housing and transportation. It's even more convenient when it comes to creation, just open your mouth, whether it's a design drawing or a speech, it's easy to get it, and if you're not satisfied, you can let it help you change it at any time.
From the point of view of the law of product development and update, Honor MagicOS 80 is just the beginning, and I believe that there will be more revolutionary technology applications integrated into personal electronic products in the future.
Honor MagicOS 80 undoubtedly brings us a vision of the future of technology, maybe one day in the near future, everyone will have a smart butler like Jarvis in the "Iron Man" movie, and this smart butler can be integrated on mobile phones, can also be integrated into watches, headphones, necklaces, rings and other wearable devices.
If you say Honor MagicOS 8The release of 0 provides us with a new personal assistant experience, like a caring housekeeper. Then Apple's Vision Pro has opened a door to the virtual world for us, bringing an unprecedented navigation experience. According to Apple's latest release, Vision Pro will be available in the United States on February 2, and pre-orders will be available on January 19. This pinnacle of seven years of development will usher in the era of spatial computing.
So, what exactly is Vision Pro? What can it do? According to Tim Cook, CEO of Apple, the Vision Pro has made us no longer bound by the display. The Mac introduced us to the concept of personal computing, the iPhone brought us the convenience of mobile computing, and the Apple Vision Pro ushered us into a new era of spatial computing.
The Apple Vision Pro is the culmination of Apple's decades of experience in high-performance, mobile, and wearable device design. It not only inherits Apple's fine tradition, but also innovates and breaks through technology and design, providing us with a new computing experience. Whether at work or in life, Apple Vision Pro can provide us with great support and assistance.
As a smart wearable device, Vision Pro is significantly different from other similar products on the market. When you put on Vision Pro, you don't feel disconnected from the world, but rather have a deeper interactive experience. The Vision Pro headset is unique in that it comes with an outward-facing display.
The display captures and displays the user's eye movements and facial expressions through the EyeSight system. This means that when a user browses content through Vision Pro, a halo flashes on the display, signaling to those around them that the user is immersed in an augmented reality (AR) world.
What's even more user-friendly is that when any person or object comes into the user's line of sight, Vision Pro automatically focuses it, allowing users to notice changes in their surroundings in a timely manner. This design not only ensures the user's immersive experience in the virtual world, but also takes into account the user's safety in the real world.
In order to work perfectly with Vision Pro, Apple has also released a new operating system, VisionOS. This is the first spatial operating system launched by Apple, and its appearance marks another innovation in the field of operating systems by Apple. VisionOS not only inherits the excellent features of macOS, iOS and iPadOS, but also provides users with a powerful spatial experience on this basis. This is all thanks to VisionOS' new 3D interface design, which enables users to see and feel digital content intuitively in the physical world.
Even better, VisionOS dynamically responds to natural light and casts shadows, a design that allows users to better understand the proportions and distances of objects. Whether at work or in life, VisionOS provides users with an unprecedented spatial experience. VisionOS is the product of Apple's thoughtful and well-designed approach to the future computing experience, and together with Vision Pro, it will lead us into a whole new digital world.
In order to realize the user's navigation needs and interact with the spatial content, Apple Vision Pro has introduced a new input system, which is designed to make the user's operation more intuitive and convenient. Controlled by eyes, gestures, and voices, the system allows users to navigate through applications with simple gaze, gentle finger tapping, or voice commands, allowing your world to expand infinitely without the limitations of the physical world. Wherever you want to go, you'll be presented in no time.
Vision Pro has a wide range of use cases, not only for entertainment and office, but also for a more immersive experience when combined with Apple's other products. For example, users can browse the library in a more immersive way with the Vision Pro, an experience that no other device can match. Especially when it comes to browsing panoramas**, the experience brought by Vision Pro is revolutionary. It can take the user back to the specific scene when the shooting** was taken, making the user feel as if they were there. This is something that Apple's other devices have not been able to do before, and it is also a major innovation of the Vision Pro.
When it comes to Apple Store, Apple mobile phone users are no strangers, because whether they are buying new products or seeking product support, they need to do so on the Apple Store. Therefore, the Apple Store is a platform that offers Apple products and services. So what is GPT Store?
From the perspective of the composition of the GPT Store interface, it is very similar to Apple's App Store, and the categories include:
lfeatured: featured apps of the week;
ltrending: the most popular GPTS in the community;
LBY ChatGPT: GPTS created by the ChatGPT team.
On January 10th, OpenAI's app store, GPT Store, was officially launched, with columns such as categories, trends, weekly picks, and more, and GPTS was also divided into categories such as "writing", "efficiency", "research and analysis", "programming", "education", and "lifestyle" according to the purpose of the app.
OpenAI Greg Brockman said this is the first step in building your own ChatGPT. The product is still in the experimental phase, but it is hoped that it will be rolled out more widely in the coming weeks. OpenAI will also highlight useful and impactful featured GPTs on a weekly basis.
OpenAI also announced a new plan to share revenue with GPT's creators in the first quarter of this year. At the heart of this program is the idea that GPT creators will be compensated based on how much users interact with the chatbot. That is, the creator of each GPT app can become a partner of OpenAI GPT Store, and can create interesting apps on GPT Store, and as long as users interact with these apps, the creator has the opportunity to share the corresponding fees.
However, OpenAI has yet to disclose the specifics of this plan. For example, it's unclear how they're going to calculate payments or how they're going to measure user engagement. If a user only tries it out for a few seconds and then closes the chatbot because they don't like it, does that happen count towards GPT engagement? Is this the only way to measure user engagement?
These are the concerns of GPT creators, and everyone is looking forward to OpenAI releasing more information soon so that creators can better understand this plan. This program is undoubtedly an encouragement to the creators of GPT and a recognition of their contributions. Everyone is looking forward to seeing how this plan is implemented and how it will affect the development of GPT and chatbots.
The GPT Store is serving ChatGPT Plus users, business users, and the newly launched ChatGPT Team. It can be seen that OpenAI not only absorbs a large number of individual users, but also targets enterprise users, and it is believed that it will not be long before many individuals and enterprises who start businesses on GPT Store will appear.
ChatGPT Team is a paid version of ChatGPT designed for small teams of about 150 people. Similar to ChatGPT Enterprise, ChatGPT Team users will be able to use GPT-4, DALL-E 3, and OpenAI's advanced data analytics features and exercise control over their data. OpenAI has made it clear that ChatGPT Team's data and conversations will not be used to train any of its models.
Additionally, users of ChatGPT Team can also create custom GPTS based on their team's specific needs, or choose to use other GPTs from the store. This provides great flexibility and convenience for the team.
As for ChatGPT Team's charging scale, the annual billing method is $25 per user per month; If you choose monthly billing, you'll need to pay $30 per user per month. This flexible billing method is designed to meet the needs of different teams, allowing them to choose the most appropriate billing method for their budget and needs. ChatGPT Team is a powerful and flexible tool designed to help teams make better use of GPT technology and be more productive and innovative.
From the development of large models to AI agents, the development timeline of artificial intelligence has reached the eve of the emergence of embodied artificial intelligence, which will have a huge impact on the way of life of human beings.
1. Single-agent application:Single-agent applications are an important application area of AI Agent, specifically, AI Agent can be used as a personal assistant to help users handle daily tasks and repetitive tasks. They are able to analyze, plan, and solve problems independently, reducing personal work pressure and improving task solving efficiency.
AI agent can help users manage daily tasks such as setting reminders, scheduling, sending emails, etc. They can automatically adjust the priority and timing of tasks based on the user's needs and habits. It helps users retrieve and analyze information. Relevant information can be found in large amounts of data to help users make decisions. Help users automate some repetitive tasks such as data entry, file management, etc. This saves the user's time and gives them more time to work on more important things.
AI agent can help users solve problems. They can analyze the problem, come up with a solution, and even execute the solution directly. AI agents can also learn from users' behaviors and preferences, gradually adapt to users' needs, and provide more personalized services. Since AI Agent is a cloud-based service, they can serve users anytime, anywhere.
2. Multi-agent system:A multi-agent system is a system consisting of multiple AI agents that can interact with each other in a collaborative or competitive manner. This type of interaction allows them to progress through teamwork or adversarial interactions.
In the collaboration mode, multiple AI agents can form a team to share information and resources to solve problems together. For example, they can collaborate on complex tasks such as search and rescue, logistics or gaming. In this mode, the AI agent needs to have good communication and coordination skills in order to be most effective in the team.
In the competitive mode, AI agents compete with each other to achieve their respective goals. For example, in some strategy games, AI agents need to defeat their opponents through adversarial interactions. In this model, AI agents need to have strong strategy and decision-making capabilities in order to gain an edge over the competition.
Whether it's collaboration or competition, AI agents can learn and progress through these interactions. They can gain new knowledge from each interaction and improve their strategies to perform better on future tasks. This capability makes AI Agent have a wide range of applications in many fields, such as machine learning, game theory, and robotics.
Multi-agent systems are an important research direction in the field of AI, which provides the possibility for us to understand and design more complex and intelligent AI systems by simulating and studying the interactive behavior of multiple AI agents. Not only does this system help us solve more complex problems, but it also provides us with deeper insights into how to design and manage a system of multiple agents. This is of great significance to promote the development and application of AI technology.
3. Human-machine cooperationHuman-robot cooperation is an important application of AI agents, which can enable AI agents to interact with human users to provide assistance and perform tasks more efficiently and safely.
AI agents can interact with human users in a variety of ways. For example, they can receive and understand a user's instructions through voice, text, or images. They can also provide feedback to users in these ways to help them understand the status and behavior of the AI agent.
AI agents can provide various types of assistance to human users. For example, they can help users search for information, solve problems, learn new skills, and even navigate complex environments. They can also provide safety when needed, such as controlling the vehicle's movement in a self-driving car to keep passengers safe.
AI agents can understand human intentions and adjust their behavior accordingly. They can learn from their behavior patterns and preferences to better meet their needs. For example, an AI agent might learn that the user likes to listen to the news in the morning, so it automatically reports the news every morning.
Human-machine cooperation enables AI agents to better serve humans, improving our quality of life and work efficiency. At the same time, it also provides an opportunity for AI agents to learn and improve, allowing them to continuously improve their capabilities and performance. This partnership is based on mutual understanding and trust, and requires AI agents to be highly adaptable and flexible. Only in this way can AI agents truly become our right-hand man in our lives and work.
4. Professional Field:AI agents can be trained and specialized for specific domains, such as software development, scientific research, or other industry-specific tasks. They can provide expertise and support in these areas, taking advantage of the pre-training of large-scale corpora and the ability to generalize to new tasks.
When it comes to software development, AI agents can be trained to understand and generate**, helping developers write and debug programs more efficiently. They can provide suggestions, autocomplete snippets, and even help detect and fix errors in them. In addition, AI Agent can help developers understand complex libraries, allowing them to become familiar with new environments faster by providing an overview of their structure and functionality.
In the field of scientific research, AI agents can be trained to understand and generate scientific literature, helping researchers keep track of the latest research progress. They can automatically extract key information from a large amount of scientific literature, such as experimental results, research methods, and conclusions. In addition, AI agent can also help researchers design and execute experiments, enabling them to conduct scientific research more effectively by providing recommendations on experimental design and experimental results.
AI agents can be trained to perform a variety of industry-specific tasks. For example, in the medical industry, AI agents can help doctors diagnose diseases, provide recommendations, and even develop trends in diseases. In the financial industry, AI agents can help analyze market trends and even help develop investment strategies.
The training and specialization of these AI agents is based on the pre-training of a large-scale corpus and the ability to generalize to new tasks. They can learn from large amounts of data and then apply that knowledge to new tasks. This ability allows them to provide expertise and support in a variety of areas of expertise to help humans accomplish tasks more efficiently and accurately. The professional application of AI agent is an important direction for the development of artificial intelligence, which will greatly promote the progress of all walks of life.
From the battle of 100 models of large models to the competition of AI agents, the past year and the year to come are destined to be a strong stroke in the development of AI.