Attracted the investment of the Jobs family twice, Brain.ai tried to invent the next generation computer interface.

[Hunting Cloud Network (WeChat: ilieyun) Beijing] reported on March 10 (Text/Fei Wang)

In December 2015, Brain, LLC(Brain.ai, hereinafter referred to as Brain), an artificial intelligence company, was quietly born in Silicon Valley, USA. Its founder was Jerry Yue, a 23-year-old serial entrepreneur.

At the beginning of its establishment, Jerry set up a founding team composed of six or seven famous doctors. Different from the route of most artificial intelligence companies, Brain is committed to exploring the technical path of establishing a more general artificial intelligence model.

In 2021, Natural, an application product based on Brain’s self-developed AI model, was launched in App Store. This is the world’s first customer-oriented generative interactive interface, which is dedicated to changing the existing human-computer interaction mode and making it more natural. With Natural, you just need to simply state your needs, and it automatically organizes the appropriate apps. Instead of looking for apps, apps come to you.

Only half a year after it was officially launched, Natural ranked in the top ten of the App Store’s list of downloads of similar applications. By February, 2022, Natural had made 8 million monthly user purchase requests in North America, and achieved an average monthly growth rate of 200%.

The Jobs family invested twice, and Lesheng impressed Laurene with Empathetic AI.

Jerry15 was admitted to the junior class of Stanford University at the age of 15 to study mathematics. At the age of 18, he founded American Small Passport Company as CEO. Before founding Brain, Jerry was the CTO of LifeNet, and founded several technology companies including billions of dollars.

Brain founder le sheng, Jerry Yue

According to reports, Jerry’s mother is an English literature professor and his father is a physicist, which makes him more involved in humanistic thinking besides technical rationality. He defined Brain as a "company standing at the crossroads of science and art".

In Jerry’s view, technology serves people. Besides the breakthrough in technical ability, interaction is another indispensable element. Just as computers entered thousands of households after the invention of the keyboard and mouse, the birth of the iPhone brought the world into the era of mobile Internet. By analogy, today’s AI is just like the computer before the invention of mouse and keyboard, which can only serve professional and technical personnel. Jerry believes that "AI interaction is an area that deserves to be optimized thousands of times."

In 2016, Brain made an important breakthrough in large-scale language models and invented the world’s first relatively universal artificial intelligence. Well-known American media followed the reports one after another, and with the recommendation of friends, it took Jerry less than two hours to make Laurene Powell Jobs, the widow of Apple founder Steve Jobs, determine his investment intention.

According to Jerry’s memory, the first time I met Laurene was in San Mateo’s office. "I simply went with a sharing mentality." Although he knew Laurene was in charge of the foundation of the Jobs family, he didn’t think about it at all.

Prior to this, Laurene had little interest in the technology circle and never directly invested in any technology company. Moreover, Brain’s AI model is just a demo, which has only been verified in theory, and there is not even a decent business plan.

In June 2016, just half a year after its establishment, Laurene officially invested 10 million US dollars in Brain, and the company was renamed Brain Technologies, Inc Brain became the first technology company directly supported by the Jobs family.

Jerry thinks that what really touches Laurene is just one word: Empathetic AI-perhaps many people have this ideal, but at the same time, it is also the goal that 99% commercial companies dare not pursue.

Laurene didn’t impose any business requirements on Brain, and didn’t even replace the equity immediately. The only condition she offered was that Jerry should not change his direction for short-term business goals and avoid the possibility of capital interfering with technology as much as possible.

From 2016 to 2020, the Brain team focused on technology research and development. In the end, the model realized that the AI model "self-taught" more than 4 million functions without human participation, and its self-generation ability entered an exponential evolution state. Among them, a set of relatively mature models launched by Brain in 2017, even after six years of hot development of artificial intelligence, this model is still difficult to rival.

The evolution speed of the model even shocked Jerry. In a test, Jerry once asked AI a philosophical question: Do people have free will? AI gives an answer beyond many human understanding: "The answer depends on the question you are discussing. The closer you are to the desire to live, the lower your free will is. If you are farther away from the desire to live, the higher your free will is. -this is an answer that is not available on the Internet. It is self-generated by AI, and it took less than 24 hours to retrain this Chinese model.

So in June, 2021, Brain completed another $50 million financing, and Laurene continued to follow suit. Other investors included Goodwater Capital, Scott Cook and WTT Investment.

It is worth mentioning that, under Laurene’s introduction, several executives of Apple at that time also began to have a strong interest in Brain. They became Jerry’s friends and saw another possibility of Siri from Brain.

Brain tried to invent the next generation computer interface, and let the App find users.

In recent decades, computers have changed all aspects of human life. From the command line to the mouse /GUI (graphical interface operating system), from the browser to the iPhone, the definition of computers and the methods of interacting with them have all changed dramatically. These changes and revolutions have one thing in common: they are the products of a great product that can simplify a revolutionary technology. However, this evolution has come to an abrupt end in the past 15 years.

Jerry said, "Since the release of the first generation iPhone, there is still no new product that can replace its status as the’ most advanced contemporary computer’. We can’t help asking, what should the next generation of computers look like? How to interact with humans? In the past five years, Brain has been trying to answer this question by combining design innovation with machine learning R&D. "

Development course of human-computer interaction

Through more general AI technology, Brain makes technology and code warm and empathetic, and finally establishes a human-centric digital world. In order to achieve the goal of universal AI, Brain chose the technical route of One-shot learning. Compared with supervised learning driven by a lot of labeled data, single-sample learning enables machines to deeply understand semantics and establish semantic representation through large-scale models under unsupervised learning training, and can learn deep relationships and new concepts by analogy only by looking at a very small amount of data.

For example, if supervised learning is a way for babies to learn-if you want to teach a baby to recognize cats, you need to show him a lot of photos of all kinds of cats. Then single-sample learning is closer to the way of adults’ cognitive world-based on prior knowledge, we can trigger strong self-learning ability with a very small amount of information, and learn or understand problems explosively.

Most of the data of supervised learning and training process are relatively independent, while the training data of human brain are relatively universal. Humans can use the learning of one thing, cognition and skills in the learning of another new cognition and skill, thus greatly reducing the cost of learning and the amount of data needed. This ability is called Transfer Learning in computer science. In the early days, almost all the research and development efforts of Brain focused on the research and development of this universal learning ability. In 2016 and 2017, Brain made some major breakthroughs and applied for core patents in hundreds of related fields.

In 2020, Brain invented and launched the world’s first instant generating interface that can help you think. It will instantly render and generate related interfaces and applications based on users’ goals, and help users plan and complete various tasks directly and personally. From then on, it is no longer you who go to the App, but the App comes to you.

Natural’s generative interface

Take the transaction application as an example. If you want to book a plane ticket, you don’t need to think about which App to go to or jump back and forth between apps to compare prices. Just say the place you want to go or the description of the destination you expect, and Brain’s interface will instantly generate relevant interfaces and search results of the whole network for you to book tickets with one click of Apple Pay.

If you want to buy fresh food, you can describe the dishes you want to cook. The search results and recommended recipes of all ingredients on different fresh platforms will be generated in front of you immediately, so you no longer need to think about every ingredient and search them one by one.

Natural query recipes

If you need to buy a gift, you can even roughly describe your friend’s personality or preferences, or even the topics you talked about recently. The generated interface will help you imagine dozens of gift ideas and just the right reasons for recommendation, and even help you write a card and send it directly.

Natural online shopping interaction

In addition, work applications have become extremely intelligent. After receiving a legal document, the email App can directly help you explain the terms, and you even ask it, "How can the terms become more beneficial to the company?" At this time, the interface will automatically call the legal AI, seamlessly generate rigorous legal advice, and automatically reply to the email after changing the legal language according to your feedback. PPT can brainstorm with you and help you complete your business plan, stock trading software can help you analyze your investment strategy, and code editor can even help you write complete code.

Brain’s second step is to make every App smart.

In order to realize the interconnection between Apps, Jerry explained to Hunting Cloud, "Suppose an app has an API, and then there are many API documents in this API. This API is a direct interface, and engineers can use different functions in different apps by calling these APIs. Brain has a core technology called’ New Kans Generation Module’, which simply means to create a tool that learns to use all tools. Our company’s natural language model can directly generate relevant codes, and based on the data of these API documents, directly generate API codes for direct access. This also makes Natural’s work all automatic. At the same time, it can generate code and use it in different ways of using this more general model, so that this AI can automatically use other tools. "

Jerry further explained, "Natural is an operating system level, based on open source. After agreeing to some protocols, it can call the contents of other apps. Then from a technical point of view, other apps can’t be rejected unless it refuses everyone’s use. "

And this is the core of Brain’s first step plan, let the App find the user, not the user to find the App. Jerry told Hunting Cloud, "Brain’s second step is to make every App smart. Netease and Baidu have the ability to process files. For example, if someone sends you an email and sends a PPT in Baidu’s network disk, you can say, can you help me simplify pages 3 and 6 of this PPT a little? What we want to do is to create the next dimension of products and create the next dimension that developers can imagine infinitely. Then this dimension is the wisdom dimension. "

Up to now, Brain has a team of nearly 300 people in China, America and Europe. The chief strategy officer is MC Hammer, the originator of hip-hop music and a well-known investor in Silicon Valley. In 2017, visual design coffee Gleb Kuznetsov joined the company. In addition, there are many AI "prodigies" admitted to Caltech at the age of 12, and many art engineers, such as the inventor of sandbox technology currently used in billions of mobile phones around the world.

Brain’s general AI model has unlimited imagination in the future. In order to realize the second step, Jerry also hopes that more and more China developers can join the team to jointly promote the changes in the history of human science and technology. At present, Brain has set up offices in Xi ‘an, Nanjing and Chengdu.