2021 was a year of great social changes (AI), such as the spread of the new coronavirus and the increased attention to DX.
Continuing from last year, this time, we asked people who are active in the front lines of the AI industry about trends in the AI industry in 2021, news that caught their attention, and AI trends in 2022 using a questionnaire. .
How do celebrities active in the industry see trends in the AI field in 2021 and trends in 2022?
Table of contents
- Mr. Ryuji Noguchi, Japan Deep Learning Association
- Please tell us the most interesting news related to AI in 2021.
- What changes do you see in the AI space in 2021?
- In 2021, how did your company’s AI service make a leap forward?
- How do you see AI trends changing in 2022?
- Detroit Tohmatsu Group Mr. Masaya Mori
- Please tell us the most interesting news related to AI in 2021.
- What changes do you see in the AI space in 2021?
- In 2021, how did your company’s AI service make a leap forward?
- How do you see AI trends changing in 2022?
- Cyber Agent Mr. Kota Yamaguchi
- Please tell us the most interesting news related to AI in 2021.
- What changes do you see in the AI space in 2021?
- In 2021, how did your company’s AI service make a leap forward?
- How do you see AI trends changing in 2022?
- Mr. Takashi Takegawa, zero to one Co., Ltd.
- Please tell us the most interesting news related to AI in 2021.
- What changes do you see in the AI space in 2021?
- In 2021, how did your company’s AI service make a leap forward?
- How do you see AI trends changing in 2022?
- Mr. Yoshifumi Tamoto, CEO of MatrixFlow Co., Ltd.
- Please tell us the most interesting news related to AI in 2021.
- What changes do you see in the AI space in 2021?
- In 2021, how did your company’s AI service make a leap forward?
- How do you see AI trends changing in 2022?
- Mr. Yoji Kiyota, LIFULL AI Strategy Office
- Please tell us the most interesting news related to AI in 2021.
- What changes do you see in the AI space in 2021?
- In 2021, how did your company’s AI service make a leap forward?
- How do you see AI trends changing in 2022?
- Accenture K.K. Mr. Hoshinase
- Please tell us the most interesting news related to AI in 2021.
- What changes do you see in the AI space in 2021?
- In 2021, how did your company’s AI service make a leap forward?
- How do you see AI trends changing in 2022?
- Mr. Akihiko Ishikawa, Aidemy Co., Ltd.
- Please tell us the most interesting news related to AI in 2021.
- What changes do you see in the AI space in 2021?
- In 2021, how did your company’s AI service make a leap forward?
- How do you see AI trends changing in 2022?
- Mr. Ryo Shimizu, Gillia Co., Ltd.
- Please tell us the most interesting news related to AI in 2021.
- What changes do you see in the AI space in 2021?
- In 2021, how did your company’s AI service make a leap forward?
- How do you see AI trends changing in 2022?
- Rikkyo University Graduate School of Artificial Intelligence Science Yoichiro Miyake
- Please tell us the most interesting news related to AI in 2021.
- What changes do you see in the AI space in 2021?
- In 2021, how did your company’s AI service make a leap forward?
- How do you see AI trends changing in 2022?
- Chisato Kunimoto, Director and COO of Kyun Co., Ltd.
- Please tell us the most interesting news related to AI in 2021.
- What changes do you see in the AI space in 2021?
- In 2021, how did your company’s AI service make a leap forward?
- How do you see AI trends changing in 2022?
- summary
Mr. Ryuji Noguchi, Japan Deep Learning Association
Please tell us the most interesting news related to AI in 2021.
| Yahoo, 8,000 employees in the entire company converted to advanced IT personnel by retraining |
What changes do you see in the AI space in 2021?
| Paradigm shift due to the emergence of large-scale language models |
In 2021, how did your company’s AI service make a leap forward?
| As JDLA, we have delivered AI education to more people under the slogan AI for Everyone. |
How do you see AI trends changing in 2022?
| I believe that we will enter the year when social implementation of large-scale language models will be accelerated. |
Detroit Tohmatsu Group Mr. Masaya Mori
Please tell us the most interesting news related to AI in 2021.
| In 2021, I was excited about the further development of AI models, such as the application of Transformer to Computer Vision and CLIP, which connects images and text. I was curious about the news related to reduction and optimization. Specifically, the RETRO “Retrieval-Enhanced Transformer”, which is 1/25th the size of DeepMind. |
What changes do you see in the AI space in 2021?
| I think that the trend toward a comprehensive AI architecture, such as combining with symbol manipulation, has grown steadily. I believe that AI research will continue to make remarkable progress. |
In 2021, how did your company’s AI service make a leap forward?
| First, with the launch of the Deloitte AI Institute announced in June this year, the AI-related services of the Deloitte Tohmatsu Group have been linked and expanded. From strategy and governance formulation, AI roadmap creation, PoC planning and execution, implementation of specific analytics, development and introduction of various specific applications such as fraud detection, demand forecasting, social listening, personalization, and customer contact support. , construction of platforms including data management, maintenance and management of data, introduction of AutoML, implementation of MLOps, training of AI and data-related human resources, support for building AI utilization organizations, etc. rice field. On top of that, Scaling AI (AI expansion and mass production), Trustworthy AI (development, management and operation of trusted AI services), and based on these, AI Fueled Organization Consulting to evolve the company by utilizing AI. offerings are also available. |
How do you see AI trends changing in 2022?
| Regarding the technical part, I think that interest in the combination of the Metaverse and AI will increase. There is a trend to promote understanding of the world through the combination of symbol manipulation and machine learning, and I think that there will naturally be a movement to accelerate this by combining it with the metaverse (or digital twin). It seems that progress in various applications can be seen in a movement that can be said to be a synergy between simulation and machine learning.This year also marks the 50th anniversary of the publication of the “Limits to Growth” by the Club of Rome, which was established in 1972 by intellectuals from around the world, stating that human growth will reach its limit within 100 years. Partly because of this, I believe that interest in environmental pollution and climate change will grow significantly. I think that the potential of AI in predicting changes in the environment and climate, optimizing and reducing energy resources to prepare for such changes, searching for new materials and developing products, etc., will attract attention. and machine learning will be important. |
Cyber Agent Mr. Kota Yamaguchi
Please tell us the most interesting news related to AI in 2021.
| I think the biggest news was the potential for general-purpose AI through pre-learning, represented by CLIP[1]. By building a machine learning model from a huge corpus in advance, it is possible to perform image recognition even without training data for a specific task, and to use text such as CLIP+VQGAN[2] and DALL-E[3]. It has been demonstrated that tasks such as generating an image using as input can be performed with higher quality than before. |
What changes do you see in the AI space in 2021?
| Socially, as seen in the case of Zillow, a US real estate tech that failed to utilize AI under the influence of COVID-19, AI-based services are vulnerable to sudden changes in data when there are major changes in social conditions. I think it was also a year when the weaknesses of |
In 2021, how did your company’s AI service make a leap forward?
| The CyberAgent AI Lab will continue to strengthen its research and development in 2021, and has engaged in many interdisciplinary studies that transcend academic disciplines. As a result, 51 research papers were accepted for international conferences and journals, and the number of accepted papers increased by 46% from the previous year. In our digital advertising-related business, we are able to provide customers with high advertising effectiveness by utilizing AI, and we are also working on technology areas such as computer graphics and voice synthesis using AI, and utilizing AI in the administrative area. is progressing. |
How do you see AI trends changing in 2022?
| AI research trends are expected to continue in 2021 with the construction of general-purpose pre-learning models and the integration of multiple technical fields. While the threshold for research and development that cuts across technical fields such as texts, images, and voices has been lowered, the disparity between those who have large amounts of data to handle them and those who do not has widened. may be lost. Also, I think we are in a phase where we should take into consideration the safety of data-dependent AI so that it does not behave unintentionally. |
Mr. Takashi Takegawa, zero to one Co., Ltd.
Please tell us the most interesting news related to AI in 2021.
| In the field of “education”, which is the core of our zero to one business, the first thing that comes to mind is Prime Minister Kishida’s news of “400 billion yen to invest in people over three years,” including the digital field. . At zero to one, we have already seen an increase in AI and digital human resource development projects led by local governments such as Kochi Prefecture and Sendai City since this fiscal year. I am very much looking forward to speeding things up. |
What changes do you see in the AI space in 2021?
| Among our customers, major companies, for example, the human resources they want to develop are overwhelmingly called “digital human resources” instead of “AI human resources”. What used to be a story about educating a select group of people for “special skills” has changed dramatically to a story about educating everyone broadly about “skills and knowledge that should be possessed as literacy.” Did.Coupled with the aforementioned government move, I feel that this trend will further intensify in 2022. |
In 2021, how did your company’s AI service make a leap forward?
| In October 2021, we released a new online teaching material titled “Autonomous Driving System Construction Complete Course”, but there are three big “leaps” hidden here.The first is the establishment of a teaching material development model through industry-academia collaboration. This teaching material was jointly developed by Nagoya University Future Society Co-Creation Organization, Tier IV Co., Ltd., and our zero to one company, and achieved release in a short period of time by leveraging each other’s strengths. Considering the fact that it has been certified by the Ministry of Economy, Trade and Industry as a “Fourth Industrial Revolution Skill Acquisition Course”, it may be said that it is an “industry-government-academia collaboration” model.
The second is the provision of a practice environment in which the GPU can be turned on and off on the cloud. This is an environment developed independently by zero to one, and in developing and providing online teaching materials related to AI and data in the future, we can deliver a new added value of providing practical cloud exercises to students. will be
The third is the external provision of our proprietary zero to one learning management system (LMS) itself. While establishing the above-mentioned new teaching material development model this time, we enhanced the function so that partner companies can enter the zero to one LMS themselves and create teaching materials.
Based on these “leaps”, we would like to further focus on developing and providing educational programs that “draw out the power to live actively with society.” |
How do you see AI trends changing in 2022?
| Just as AI itself is entering a phase where it is required to be used in actual business, I believe that AI and digital human resources will also be the year when attention will be focused on their “utilization”.For cutting-edge human resources with more specialized skills, I think that a major revision of the personnel system itself, such as job responsibilities and treatment, will be required. From the perspective of how to improve “digital skills”, I think that it will be necessary to revise and improve the education and training system. In any case, if you seek “perfection”, you will lose a year or two immediately, so regardless of the size of the company, it is important to proceed quickly, assuming that there may be some failures. I think so, and I would like to support them by sharing best practices as a zero to one initiative. |
Mr. Yoshifumi Tamoto, CEO of MatrixFlow Co., Ltd.
Please tell us the most interesting news related to AI in 2021.
| This is “DALL E” from Text-to-Image generation. If you enter text such as “avocado-shaped armchair”, it is a model that generates an image that matches it. I was really surprised with such a picture. In addition, models such as “GLIDE”, which has been further developed, have also appeared and are showing excitement. Not only is it interesting, but it is also very promising for business applications such as similarity calculation between text and images. |
What changes do you see in the AI space in 2021?
| In 2021, an unusual competition was held to improve the accuracy of machine learning by improving the dataset, and Andrew Ng took the lead in recognizing the importance of the quality of the dataset. AI is a very important idea when it comes to be used in various industries and responds to specific applications, and the emergence of tools that support Data-Centric ML development is expected. |
In 2021, how did your company’s AI service make a leap forward?
There are also good examples from customers who have started using the no-code AI utilization tool “MatrixFlow”.
We will continue to devote ourselves to the democratization of AI so that many people can use “MatrixFlow”! |
How do you see AI trends changing in 2022?
| With the advent of the Metaverse, I think the use of machine learning in VR will become more popular this year. When it comes to the use of AI in cyberspace, until now, most of the applications have been for business in the form of digital twins and AI, but cyberspace AI for consumers will become popular. |
Mr. Yoji Kiyota, LIFULL AI Strategy Office
Please tell us the most interesting news related to AI in 2021.
| I was greatly shocked by the news of the death of my university teacher, Professor Makoto Nagao, who was known as a pioneer in the field of artificial intelligence in Japan.NHK News: Kyoto University Professor Emeritus Makoto Nagao dies 84 years old Development of machine translation system etc. (May 26, 2021)
Nihon Keizai Shimbun: Makoto Nagao passes
away Former President of Kyoto University (May 26, 2021) I think that Professor Nagao’s research style of “going next” has many hints for further development of AI in the future. |
What changes do you see in the AI space in 2021?
In the business field as well, I feel that it is becoming possible to have down-to-earth discussions based on realistic applicability, such as what machine learning can and cannot do.
On the other hand, I think that the flip side of the growing impact of AI systems on society is the growing ethical risk. |
In 2021, how did your company’s AI service make a leap forward?
| The floor plan analysis project, whose conception started around 2015, has come to fruition as a new service called “LIFULL HOME’S 3D Floor Plan” over the course of six years.In the future, we would like to further advance AI research and development to create groundbreaking user experiences. |
How do you see AI trends changing in 2022?
As social implementation of AI progresses further, I think that social acceptability issues, such as whether it is acceptable to acquire such data, will further determine the success or failure of AI services.
In order to deal with these problems, I think we will need a more comprehensive perspective that includes the humanities. |
Accenture K.K. Mr. Hoshinase
Please tell us the most interesting news related to AI in 2021.
| There were some noteworthy news, such as the start of the Digital Agency and Facebook becoming Meta overseas, but personally, it was a year that paid attention to overseas movements related to responsible AI. In the EU, a resolution was adopted urging police to ban the use of AI-based facial recognition and crime prediction, while in China the installation of surveillance cameras is progressing, and by linking surveillance cameras and AI, crime can be deterred. I feel that it was a year in which different trends in the world, such as connecting, were visualized. |
What changes do you see in the AI space in 2021?
| As the novelty of technology has faded, the question of what kind of business results the introduction of AI has led to has become more important. While the democratization of AI has accelerated, it is precisely because AI has become a commodity that we feel that users are becoming more distrustful of AI. It was a year when the difference between the work that was not so became clearer. I think this difference will become apparent in the form of corporate performance this year. |
In 2021, how did your company’s AI service make a leap forward?
We were able to evolve the service in a way that commits to business results. For example, when introducing an AI solution that supports customer contact points, there is a need to commit to specific results such as integration with UX, call volume reduction, cost reduction, CS improvement, and customer acquisition, rather than simply introducing an AI system. In response, Accenture has strengthened its comprehensive support system that integrates knowledge in the areas of customer experience, operations, AI, and technology.
In addition to creating business results with AI services, we are also promoting collaboration with universities and medical research institutes in order to address social issues beyond that. |
How do you see AI trends changing in 2022?
| As the rarity of technology has faded and AI has become a commodity, I expect that the question will be more about management impact than technological superiority, and the creation of new value unique to AI rather than simply automating human work. As scale is required and AI will be involved in more important decision-making, mechanisms for continuous value creation, including ML Ops, and responsible AI that emphasizes reliability and responsibility will also be introduced. I guess. In Gartner’s AI hype cycle, the focus has so far been on core AI technologies such as Deep Learning, but the focus is shifting to data, ModelOps, and other aspects of business operations and responsible AI. . |
