In the movie “Oppenheimer”, Oppenheimer’s wife said this to him: The world is turning rapidly, and your moment has come. Perhaps, in the 20th century, applied physics and the Internet changed the world we live in. Now, the arrival of the era of artificial intelligence has once again pushed the course of human history onto another track. If you say that when AlphaGO defeated the Go master Lee Sedol a few years ago, you still thought that artificial intelligence (Artificial Intelligence, AI for short) was far away from our lives. And now, when more large-scale AI model products come out, everyone can truly feel its impact and impact on our lives. When the first batch of workers have begun to embrace AI as a production tool, we seem to be ill-prepared for this new trend that is emerging before us.
In 1651, the British scholar Hobbes described an “artificial animal” that could be manufactured in his book “Leviathan”. This was probably the earliest written description of AI. AI originated from human fantasy. Early AI can only complete some specific tasks that require a lot of calculations based on preset programs, such as playing chess, stitching patterns, classifying s, answering simple questions, etc. Currently, with the help of the latest technologies such as deep learning, some AI devices have achieved full automation and can independently reason and solve real-life problems, such as language translation, automatic transactions, face recognition, driverless driving, etc. The development and application of artificial intelligence technology are profoundly changing human life and will inevitably impact the existing ethics and social order. This has also made many people anxious about their future – will we lose our jobs because of AI ? People who are already using AI as an auxiliary tool have told us the answers in their own minds.
“I can’t live without it now”
Xiaomei is a well-known food blogger and a senior writer of food columns in major magazines. From her writing, we can always feel the vivid flavor of food, and novel angles that you have never thought of will burst out from her writing. For example, in “A movie about aliens, why do you need to eat braised food?” ” wrote:
The large/small intestines flowing with intestinal oil spray the aroma of oil in the mouth; the lungs covered with trachea bring a slight creaking sound to the mouth; the fire filled with old soup is salty, heavy-flavored and not refreshing, but it is still delicious. Let the stomach and brain receive the satisfaction of “fullness”.
With beads of sweat on his head, he smashed the Erguotou into his mouth. The spicy and spicy aroma slid down along with the delicious aroma of oil, making his internal organs feel hot.
When you raise your head when you are tired, you can still listen to the gods and goddesses around you telling stories and making jokes. Life becomes brighter in such a scene, so that the development and ending of human destiny can seem sparse and ordinary in this hustle and bustle. .
This is her view on the novel “The Three-Body Problem” and her attitude towards embracing AI. What Liu Cixin chose is the perspective of an ordinary Chinese scientist living in the earth’s civilization. Xiaomei’s contrast between familiarity and unfamiliarity, smallness and grandeur is a unique perspective that allows us as an audience to perceive. After all, we are familiar with braised food, and we are also familiar with life. The same is true for AI. After all, for everyone, life continues every day, and technology changes with each passing day. When we look back, we can still see how far we have come with that bowl of stew.
So when AI first became accessible to ordinary people, Xiaomei couldn’t wait to start understanding and using it. This stems from her fear of the unknown. “You know, it’s like a known and terrible enemy. I have to understand what he is first. If you want to win against your opponent, you have to become friends with your opponent. That’s how I started to gradually get in touch with him. , now I admit that I can’t live without him, and I upgraded from the free version to the paid version. In the final analysis, AI is really too efficient.”
Now Xiaomei is no longer afraid of writing in English. Before the epidemic, local media in the United States asked her to write some articles. At that time, she had no confidence in her English ability. In the Chinese context, she is a good author, but in English it is completely different. I couldn’t ask anyone for this kind of thing, so I had to let it go. “Now I’m thinking, if there had been such a complete AI tool as it is now, I would have been an internationally renowned bilingual author.” But she quickly added, “It is indeed a confidence booster . At this moment. I don’t know the future.”
Whether it is information query or translation and writing, the emergence of AI has brought great changes to Xiaomei’s life. First of all, it is a lot of data collection work. If you ask AI what this book is about, how does it compare with other books of the same type? Compared with what is the difference, AI will give you a rough summary of a few points, and also give you the differences between the two, which saves a lot of time and cost in data search. In addition, when she completes text creation, she will hand over this part of the content to AI for error correction. AI will learn personalized language habits based on the corpus fed by the user daily to adapt to her style.
In Xiaomei’s view, AI is essentially similar to the pencils, ballpoint pens, fountain pens, and watercolor pens invented in our era. Generative AI is a creative tool. And every era produces its own tools. No one has ever said that the birth of these stationery ruined painting and art, and the same goes for AI. What it can ultimately do is help people live more efficiently.
Humans and AI have a complementary relationship. AI makes up for the limitations of human knowledge, and humans also share their most valuable experiences with AI.
Opportunity or threat?
Not long ago, there was an article titled “The first batch of people to lose their jobs due to AI? The article “Game Original Artists Are Ushering in Change” has exploded on the Internet. Due to the emergence and development of AI, not long ago, a game outsourcing company laid off 30% of its employees, most of whom were original painters.
Today’s AI drawing tools can not only complete high-quality illustrations in a short time, but also shape the beauty and creativity of some s or characters. This can almost easily achieve the professional skills and experience accumulated by the original painter through long-term efforts. A retired original artist views AI this way – it seems that AI has begun to truly threaten his job.
With the explosive birth of AI products such as Midjourney , more and more industries have begun to sense a sense of crisis, and controversy and praise coexist. What can AI do? With simple communication, tasks such as drawing, coding, translation, and video editing can be completed, making it almost impossible to tell the difference between machines and humans. In fact, while the AGI (general artificial intelligence) industry is ushering in a new round of advancement, many people are worried about being replaced.
In the view of architect Yu Qian, technically speaking, AI is far from being able to replace human labor. “I used Midjourney to do some covers, and Stable Diffusion to do some scenario scenes and regional planning, but this thing is more figurative and I can’t put it into my work. But if Party A uses this to make a The requirements may be more intuitive and clear to designers.”
In Yu Qian’s eyes, AI is not as scary as imagined. “At first, I felt that there was too much to learn, because it looked so powerful and seemed to replace my job. But after you use it, you will find that it may indeed replace it, but not so quickly. It’s not even something that will happen in the next ten years.” In terms of graphic deduction alone, it is difficult to form a similar logic to humans, because AI is based on learning from big data, and users send out keywords, and the keywords must be able to Corresponding to Curry’s information, there is also an issue of understanding involved. So it seems that a sense of crisis is imminent, but the fact is that it is still far from what people want.
But AI still brings unprecedented efficiency to our work. For example, on Xiaohongshu, we can see a lot of AI-generated posters. Including a lot of analysis work, AI also brings twice the result with half the effort, such as thermal energy analysis, energy analysis, air flow analysis, etc. In the past, dozens of people may be needed to distinguish different majors. In the future, it may be possible for five or six people to complete an entire project.
For now, AI’s performance is not always reliable. “It’s like a veteran in the workplace. It will publish an edition for you regardless of whether it’s right or not. If you tell it ‘wrong’, it will publish another edition for you. Then if you say ‘wrong’, it will say ‘I’m sorry, I was wrong’.” ‘, that’s it. The same goes for an AI chatbot that became popular this year. I asked it some questions before, but it just talked nonsense. I asked it to list the contents of a few regulations for me, and it opened its mouth and came, Then I said, ‘Have you checked’? It said, ‘I’m sorry, I was wrong.'” Yu Qian said.
Practitioners in other industries are also beginning to understand the boundaries of AI. Xu Jie, who is engaged in medical translation, found that Al Translation’s translation in the medical field is similar to that of translation software on the market, with only a “passing” score. For rigorous medicine, a little fallacy may be life-threatening. Most doctors will hardly use it even if they read the literature because the error rate is too high. But we should all be soberly aware that it is gradually improving itself.
Meet AI at your fingertips
The changes brought about by AI are improvements in efficiency and the liberation of everyone’s time. Just like before, the emergence of computers and the Internet only freed up your time for mechanical labor, giving you more opportunities to create more creatively. And this is also the reason why mankind has achieved more achievements in the past hundred years than in the previous millennium.
As one of the practitioners in the AI industry, Xiao Zhao, an AI engineer at vivo, said that he is now completely unable to adapt to life without AI. For example, he relies heavily on voice assistants when using his mobile phone; he cannot do without voice interaction when driving . and assisted driving. Whether he is writing materials, doing research, or reviewing documents at work, he needs large models to help with a lot of searches and text work; he even uses large models to arrange his daily work and schedule, and this kind of “you can’t go back once you use it ” The change of “go” is truly changing Xiao Zhao’s work and life.
Although large models currently exist in the world in the form of web pages and apps, the high threshold for use makes them far from being recognized by the general public. Companies currently working on large-scale model research and development can be divided into several factions: traditional Internet giants (Google, Tencent, Baidu, etc.) , experts specializing in the field of AI (iFlytek, etc.) , hardware manufacturers (vivo, OPPO, Xiaomi, etc.) etc.) , and of course some application developers or scientific research institutions.
Among them, “mobile phones are a special existence for large models. They appear in almost all scenes in life, are closest to users, and have a mature ecosystem. This makes mobile phones the best carrier for large models. Major mobile phone manufacturers also They are all working hard on this track, and vivo is the hardest working one among them,” Xiao Zhao said.
Prior to this, AI on mobile phones has been in a state of “hidden in the market”, and more of the processing results are directly presented (such as information push algorithms for thousands of people) . In addition to functions such as voice assistants, many Few people can intuitively perceive the existence of AI. And now the emergence of large models has given AI a subversive form, intuitively appearing in front of us.
In the modern AI industry, there is a very interesting term called “emergence”. If in the past engineers taught AI step by step how to grow and how to do certain things, then now after inputting a sufficient amount of knowledge ( data) to AI , AI has begun to acquire capabilities beyond the expectations of engineers. It’s like a child learning to speak. He can only learn a few words at first, but as the knowledge he continues to acquire accumulates to a certain level, the child suddenly starts to be able to speak whole sentences, even if no one has taught him. This is called “emergence.” The growth of large models is also related to “emergence”. This seems to herald the arrival of a new era, just like the “Jarvis” when we watched Iron Man. When it comes to reality, how to shrink this cross-generational technology into the square inch screen of a mobile phone has become the direction that Xiao Zhao and even the entire vivo AI Research Institute are working hard.
As early as 2016, the emergence of AlphaGo caused a huge shock to the entire mobile phone industry. The best human chess players have been defeated by AI several times. At that time, vivo felt that AI was one of the lowest technologies among all technological innovations and the future trend, so vivo began preparations in 2017 and officially established the AI Global Research Institute in 2018. In the same year, vivo established the Graph Research Institute, whose mission is to build a knowledge graph system for AI training through a professional team. So far, the data volume of the entire graph has exceeded 2800T. It is roughly equivalent to 160 million books, 2.5 times the size of the National Library of China. This kind of rainy day preparation is actually rare among mobile phone manufacturers.
“In the words of the director of our research institute, we hope that the future mobile phone can become an “intelligent” that can perceive, recognize and even predict your intentions, then help you or even make decisions for you, and then call a third-party APP to execute Decision-making, management and execution at the same time, and finally it is up to you to judge the results of the measurement execution. In this way, after continuous learning, your mobile phone will become the “intelligent” that knows you best. And the large model can be said to be only the most important part of this entire path. It is a small basic step, but essentially it solves the problem of semantic understanding in human-computer interaction,” Xiao Zhao said.
Of course, it is not easy to present large general models on mobile phones. He is not just developing a program. “Mobile phones are seemingly common but very complex products. Facing the new field of large models, we have to consider many dimensions. In addition to our own large model capabilities, we also have to take into account cost, hardware support, functionality, etc. consumption, user data security, etc.” Xiao Zhao explained. “So after thinking about it, the final solution is that we make five large models with different parameter levels to meet the needs of different scenarios. Looking at the entire mobile phone industry, there is no other company that is as cost-effective as us. ”
“Having these large models is not the end. We need to use large models to deeply call various sub-capabilities in the entire system. It is actually a reconstruction of the mobile phone system. On the one hand, you need large models to be able to understand your demands. At the same time, being able to accurately call each sub-capability is like we have piled up a huge building block model, and we have to be able to pull out a piece of it anytime and anywhere. At the same time, we also need to open up many external applications, so that software such as taxis and takeaways can be integrated with The large model completes the handshake, so the large model made by mobile phone manufacturers is an ecosystem in itself, not a software with a single function.” Xiao Zhao said, “On our upcoming OriginOS 4, the large model self-developed by vivo AI will be launched for the first time. Matrix, there will also be some practical applications that will be shown to everyone, so you can look forward to it.”
The new era of AI may be much closer than we think, but concerns about whether AI will replace humans are also rising. “The ultimate goal of research and development is not to let AI replace people, but to let AI help people solve more arduous and transactional tasks. Technology can truly serve people and let technology release people’s irreplaceable creativity. “Xiao Zhao explained.
For everyone, AI will not have an impact on our lives. On the contrary, it will free up more of your time, because the creativity that work really requires lies in each of us. In this process, new technologies make us start to think, what is really worth doing?
The ultimate goal of technological development is to benefit all mankind. Throughout the history of technological development, humans have created new technologies and adapted to them. The development of technologies such as nuclear energy and cloning is accompanied by various concerns, but humans will eventually find the correct way to open all technologies.
The future is here, and AI that belongs to all mankind is on the way.