Image credit: Unsplash
In 2016, AlphaGo developed by DeepMind came out, completely triggering the global artificial intelligence development boom, and since then, the new generation of artificial intelligence represented by machine learning, especially deep learning, has made breakthroughs in a more advanced, complex and autonomous direction, bringing new changes and opportunities to economic and social development.
Throughout the 60 years since the birth of artificial intelligence, we have experienced the Pre-AI era, the golden age, the first trough, the second prosperity, the second trough, and we are currently in the third wave of artificial intelligence.
At the same time, the internet traffic era has turned upside down, from an incremental market to a stock market, and the demographic dividend has become a thing of the past. The duration dividend created by a number of short video platforms represented by Douyin and Kuaishou is also about to peak.
The massive data generated by the rise of the Internet, the development and change of hardware technology and Moore's Law have brought about rapid advances in computing power, which has promoted the popularization of deep learning technology in the field of artificial intelligence, and promoted the rapid development and rapid industrialization of speech recognition, image recognition and other technologies. Some people believe that the development of artificial intelligence and the release of efficiency dividends are the new second growth curve that the entire Internet is looking for.
At the national level, both the Chinese and American governments have adopted artificial intelligence as a leading strategy for the future. In 2017, artificial intelligence was written into the State Council's Government Work Report for the first time, officially entering the national strategic level. Last year, the mainland government promulgated a series of action plans and outlines to continue to promote the development of artificial intelligence from the top-level design level. In the outline of the 14th Five-Year Plan, it is clear that artificial intelligence will become one of the most powerful boosters for building an innovative country, achieving new industrialization and informatization, and promoting high-quality economic development.
According to the Tortoise Intelligence report, AI investment totaled $77.5 billion in 2021, an increase of 115% year-on-year from $36 billion last year, which is the largest year-on-year increase in AI investment in 20 years. Coinciding with the advent of the era of hard technology, many investors believe that artificial intelligence will show an optimistic trend in 2022.
This article will analyze the following three aspects of the problem based on the dialogue between Investment Montiqi and a number of investment institutional partners:
1. In 2022, what is the investment logic of artificial intelligence?
2. Where are the new opportunities for the artificial intelligence track?
3. Can Chinese Intelligence achieve overtaking in curves?
What is the investment logic of artificial intelligence in 2022?
In recent years, everyone's understanding of the basic capabilities and algorithmic data of the entire AI has been deepening. Shi Jianping, investment partner of Lanchi Venture Capital, explained the three logical points of investment in the field of artificial intelligence from the macro level to Meng Tiqi.
The first is to pay more attention to the deeper penetration of vision and voice into the industry, such as deepening the application of the industry. Artificial intelligence is an important driver of efficiency improvement, both at the tool level and at the application level. In the past, everyone's impression of AI application scenarios may still stay in face recognition, vision, etc., but Shi Jianping believes that with the continuous penetration of AI, the deepening and popularization in various industries will be greatly improved in the future. For example, in the industrial industry, artificial intelligence can provide great help in design and intelligent decision-making.
The second is that "AI+" is expected to become a new format, and the emergence of "Internet +" has had an important impact on economic development. Especially AI+ science, AI and science are two different paradigms for humans to understand the physical laws of the natural world. For example, BioMap, an innovative drug development platform driven by the biocomputing engine initiated by Baidu founder Robin Li, is an example of AI+ drug discovery.
Batu Biotech combines advanced AI technology with cutting-edge biotechnology, realizes the world's leading protein pre-training technology through high-precision AI antibody design, and establishes the ability to predict key parameters such as affinity and stability of antibody drugs. On this basis, through the active and reinforcement learning methods of dry and wet closed-loop experiments, the design and understanding process of antibody drugs is greatly accelerated. Using the hyperscale graph neural network, high-dimensional target and disease association laws are extracted from massive biological data, and the collected high-quality verification data is combined with AI technology to achieve efficient new drug discovery.
In addition, including AI + materials, there is a great role in promoting materials, Shi Jianping believes that this is an organic combination of artificial intelligence and science, using the first principles and artificial intelligence integration, doing a wider field of penetration, will continue to find new scene breakthroughs.
Third, the underlying logic of AI technology is still challenging. The development of the underlying tools of the entire AI to today, due to the high barriers of the underlying tools of AI, artificial intelligence is still facing great challenges if it wants to truly achieve popularization. However, in terms of the basic capabilities of the underlying AI, Shi Jianping believes that it will continue to iterate and evolve, including the general model, large model, high-performance computing and other fields that everyone has paid special attention to recently, providing important basic capabilities for the sustainable development of AI, and will also promote the better combination of AI and knowledge graphs to help artificial intelligence enter an era of popularization.
Refined to the project itself, Xie Zhenliang, a partner of Jiuhe Venture Capital, told Meng Tiqi that the investment in artificial intelligence is not limited to a certain subdivision track, but mainly depends on the technical content of the project itself, the combination of landing scenarios, the market space and whether the team personnel structure is "strong" and other comprehensive factors to judge.
In another way to understand, it can be considered that the major subdivisions of artificial intelligence, whether it is driverless, robotics, medical, financial or other fields, will erupt into huge social benefits in the future.
Where are the new opportunities for ai in the future?
In talking with institutions, when it comes to the future opportunities of artificial intelligence, more investors will focus on the current card points and think that solving the card points is the future opportunity. One of the most discussed issues now is the interpretability of artificial intelligence.
In November 2021, the Recommendation on the ethics of artificial intelligence, the first global ai ethics protocol adopted by the United Nations UNESCO, proposes ten AI principles, including "transparency and interpretability".
Now many AI applications behind the algorithm working mode and algorithm training data is unknown, this unknown to the operation of society is very critical, especially in the medical, automatic driving and other areas directly linked to life safety, as well as the public's more sensitive privacy issues. Explainability is also where there has not been much of a breakthrough in recent years.
In the "Interpretable Development Report 2022" published by Tencent Research Institute, the significance of 5 interpretable AI is summarized:
1. Help users enhance their confidence and trust in AI systems.
2. Prevent bias and promote algorithm fairness.
3. Meet regulatory standards or policy needs.
4. Understand and verify the output of the AI system and promote the improvement of the system design.
5. Help assess risk, robustness and vulnerability.
In other words, interpretable AI is primarily aimed at four groups: direct users, policy and regulators, scientific research, and the public. At present, interpretability has become a "hurdle" that AI urgently needs to cross.
If "AI+" is the future development trend, then new opportunities may also be seen in the places most affected by AI.
Xie Zhenliang believes that the industries most affected by the development of AI are concentrated in the early autonomous driving and automobile-related fields, and from the perspective of the overall industry, the medical industry is more affected, especially in the fields of AI-assisted pharmaceuticals and medical imaging.
Shi Jianping told Investment Montiqi that there are two parts that are most affected by artificial intelligence, the first is that the enterprises in all industries have accumulated relatively more data on digital transformation, followed by enterprises with relatively high gross profits. In fact, every industry has enterprises that meet these two preconditions, which means that each industry has new opportunities. From the perspective of the next five years, in addition to some enterprises that are easy to obtain data such as the Internet, they will also sink into some so-called traditional enterprises, and even penetrate more into the primary and secondary industries to create more opportunities and value for society.
When it comes to commercial space and maturity, the applications in the field of image and audio are more mature, including automatic driving, security, face recognition, translation, speech synthesis, etc.
Another is NLP, human-computer dialogue in a super model like Open AI, in fact, the language generation ability has reached a high level, but at present NLP is not mature enough, some logical thinking and connection of context capabilities need to be improved. However, because of immaturity, the model is still iteratively optimized, and there are greater opportunities for image and audio, and the general dialogue will be a breakthrough.
Looking to the future, can China achieve overtaking in curves?
Due to the success of the Winter Olympics, artificial intelligence has also become one of the hot topics in the recent Winter Olympics. From the Winter Olympic Sign Language broadcasts of digital people, wearable thermometers, and AI sentries in ski resorts, to AI referees and a series of intelligent robots that have become content that the whole platform has reported, friends who usually have little to deal with AI are also forwarding comments.
Although for those engaged in artificial intelligence tracks, some AI applications in the Winter Olympics may only be specially optimized for some scenarios, but after the publicity of the Winter Olympics, the public awareness of artificial intelligence has been improved, and it is not limited to Chinese citizens, with the influence of the Winter Olympics, it has also cultivated the habit of world citizens to make artificial intelligence a part of life. And this is just the beginning, and gradual penetration is also a necessary path for the development of artificial intelligence.
Relying on the establishment of artificial intelligence living habits can help artificial intelligence develop the big market of To C, and also correspond to the call for the popularization of artificial intelligence. There are many applications and scenarios in the To C field, and it can even subvert the entertainment methods of human beings, such as the hot metacosm and virtual digital people in the past year. However, in the case of immature technology, To C will be slower than To B, and the receptivity of B will be stronger. Shi Jianping believes that at present, the country is guiding to the field of hard science and technology, and for the impact of the whole society, To B may be a more value-creating thing.
From the perspective of external factors, hard scientific and technological strength has become the key point of the great power game. In terms of objective facts, compared with technologies in the field of partial engineering such as lithography machines, the gap between China and the United States in artificial intelligence, including cloud computing, is not large, and many investors have said that they are optimistic about China's future scientific research capabilities.
For such a situation, Shi Jianping told Meng Tiqi that China has quite a lot of advantages, first of all, the scene and scale can be said to be the world's largest. In terms of the contribution of a single country, China is also very leading, and China is more willing to try in the field of new things.
At the same time, in the field of artificial intelligence research, Chinese occupy half of the country, and in terms of paper publication, Mr. Huang Xutao ranks first with 1477 papers; in the statistics of the number of citations, Han Jiawei, Wu Enda, Sun Jian, He Kaiming, etc. rank in the forefront, with more than 60,000 citations respectively.
China has a rich scene and a sufficient talent pool, so Shi Jianping believes that some world-class products can emerge in China. As for whether China can achieve curve overtaking, this requires a lot of efforts, and the challenge we face now is original innovation.
Back to the investment outlook in 2022, it is still at the level of tools, infrastructure and industry penetration, as everyone's awareness continues to improve, the production value will become more and more obvious, and optimistic about the artificial intelligence industry is actually a rational behavior.
As for when a new algorithmic framework that can disrupt everything will emerge, we have no way of predicting, we can only prepare in advance for the arrival of this day on the eve of the qualitative change and outbreak of artificial intelligence.
(This article was first published on the Titanium Media APP, and the author |.) Zhang Yinan)