Report by HRI researchers
Kenichi Shikano
Senior Manager
3rd Research Department
Masayasu Watanabe
Senior Researcher
3rd Research Department
Technology Strategy Group
Since the emergence of generative AI (GenAI) technology, there has been a lot of attention on its ability to enhance efficiency. However, when considering the future, this technology also holds the potential to generate new services and business opportunities, ultimately transforming the social structure. In actuality, there are already startups that being keenly aware of the possibilities of GenAI technology and the changes that it can bring to society. They also have embarked on the development of GenAI models, tools, and applications and their implementation in society. Due to their efforts, signs of a social transformation are beginning to appear.
The rapid development of GenAI is bringing about many benefits, but problems must also be considered, that involving privacy, copyright infringement, and the generation of misinformation. There is also the possibility that this dramatic penetration of GenAI in society will make people anxious about having to adapt to social changes, losing jobs, etc.
Based on above issues, this paper is organized as follows. Section 1 will draw a broad picture of the changes in society that are expected to take place beyond 2030 as a “future mandala” and will explore how GenAI might change our world through this future mandala. Continuing on, Section 2 introduces the technologies of startups as signs of things to come, and finally, Section 3 describes in detail the risks posed by GenAI and countermeasures to those risks.
Technical innovation through GenAI has been advancing at a remarkable speed as reflected by the announcement of many large language models (LLMs) starting in 2021. In addition, the number of parameters of these LLMs has been increasing exponentially resulting in the appearance of a wide variety of models. Moreover, models specializing in specific industries, specific languages, etc. are being developed thereby making it possible to cope with the unique vocabulary and logic of individual industries and languages. As a result, the application of GenAI is now expanding in a variety of industries including finance, medicine, and education. This expansion is not limited to language-related applications—multimodal GenAI that can handle images, audio, and other media and new application fields combined with robotic technologies are entering into the picture.
Against this background, it is exceedingly important to consider what kind of change this rapid progress in GenAI technology will bring about in the society of the future. To this end, we used the “future mandala” method to visually depict the social changes that the use of GenAI might produce beyond 2030. This is a methodology that groups together possible future trends in different categories in a radial shape. The categories that we have adopted in this research are politics, economy, and the environment, society and life, and industry (Figure 1). This method clarifies the elements that will affect the future in individual fields. It also makes it easy to understand how those elements are related to each other. Through this analysis, it becomes possible to predict complex future scenarios in a systematic and multifaceted manner. In addition, combining various keywords entered in the future mandala can be used as a step toward drawing more realistic future scenarios.
Source: Hitachi Research Institute
Figure 1: Future mandala of GenAI
In this paper, we would like to formulate three future scenarios: Environmental neutrality and economic growth (1.1), Well-being in a super-aging society (1.2), and Business-model innovation via collaboration between humans and GenAI (1.3).
Based on the future mandala, there are three keywords that will play a central role in achieving both environmental neutrality and economic growth beyond 2030: “Extreme weather and disaster prediction,” “Autonomous interoperability of urban infrastructure,” and “Increasing of the electrical demand at data centers.”
In the world beyond 2030, localized extreme weather will be concerned to occur frequently due to the rise in the earth’s average temperature. Therefore, predicting such extreme weather and responding to it accordingly will be critical issues. Under those situations, GenAI will play the role of analyzing massive amounts of data (information obtained from fixed-point observations, social networking services (SNSs), etc.) and predicting extreme weather in real time.
In addition, incorporating GenAI in social infrastructures such as transportation and energy will promote information sharing and autonomous cooperation among these different systems. As each social infrastructure will come to understand each other’s status through GenAI and make appropriate adjustments in an autonomous manner, the resilience of society overall can be expected to increase.
On the other hand, the operation of GenAI will consume large amounts of power. To cope with this problem, the evolution of energy management technologies along with the development of small-scalemodels and investment in new energy sources will be indispensable. In particular, the research of new energy sources such as nuclear fusion will be the key to supporting the sustainable use of GenAI (Figure 2).
Source: Hitachi Research Institute
Figure 2: Autonomous interoperability of digital and energy infrastructures
This scenario is drown up by extracting the four keyword expressions of “Generation of drug discovery hypotheses and personalized medicine,” “Visualization of happiness,” “Community expansion using AI,” and “Personalized education” from the future mandala.
By 2030, GenAI will be analyzing complex data such as genetic information. Therefore, the life sciences will be making significant progress enabling individualized drug discovery that can extend lives. A longer lifespan will be a factor driving a shift in the focus of well-being from physical health to mental health.
In a world in which mental well-being will take on importance, GenAI will analyze data from people’s facial expressions, speech patterns, and other characteristics to identify emotions. This analysis will also include lifestyle patterns, human networks, past experiences and memories, and more. It will be possible to visualize and provide services to improve well-being based on the analysis of happiness levels of individuals and the community. In addition, the expansion of local communities and the appearance of services that personalize lifelong learning should bring individuals and society even closer to each other and improve the well-being of society overall (Figure 3).
Source: Hitachi Research Institute
Figure 3: From physical health to mental health
We selected the three keyword expressions of “Exponential increase in creativity,” “Collaborative operation between humans and GenAI,” and “Autonomous industrial robots” in investigating this scenario.
Labor shortages will become increasingly severe in the 2030s in both developed and emerging countries. To deal with this problem, GenAI will be used in three types of areas: physical space (production sites), virtual space (site management/supervision), and intellectual creative space.
First, at production sites, the introduction of autonomous industrial robots equipped with natural language processing and missing-data-completion technology will make up for labor shortages and enable flexible responses to changes in production conditions. Next, in site management and supervision, the appearance of industry-specific GenAI trained with manufacturing know-how will deal with the shortage of new employees as successors to experienced workers. This AI will support decision-making in relation to new employee training and the drafting of on-site work procedures. In addition, avatars will be generated to simulate workers, customers, and other in a metaverse space. This space will completely reproduce a worksite to enable decision-making training under conditions closer to reality. This will enable efficient and sustainable human resource development and production site management.
Finally, in intellectual creative space, GenAI will be able to quickly generate hypotheses for business strategies and proposals for new product prototypes from massive amounts of data. In this respect, GenAI will exceed human capabilities. However, it’s only humans and not AI who possess elements such as empathy, body awareness, and practical knowledge. It is therefore humans who will select what is most effective and innovative from hypotheses proposed by AI. They willcombine these hypotheses with thier own intuition and experiences thereby enhancing creativity and generating new value. This collaboration between AI and humans will have the potential of fostering business innovation while making good use of each other’s strong points (Figure 4).
Source: Hitachi Research Institute
Figure 4: Augmenting creativity by collaborating with GenAI
As described in Section 1, AI that controls social infrastructures with low power consumption in a dynamic manner plays a core role in the category of "politics, economy, and the environment." In "society and life," drug-discovery AI proposes hypotheses for extending the healthy life expectancy of individuals, while AI that can decipher complex on-site conditions plays a core role in "industry." Taking these new initiatives into account, startup companies are taking the lead in developing GenAI models, tools, and applications and implementing them in society. In this section, we introduce case studies of such startups and investigate what kind of services could be provided by applying their technical innovations.
The Small Special Model (SSM) developed by the startup company Aitomatic, Inc. in the United States is attracting attention as an industry-specific small-scale model. It is a model specialized for individual devices in actual operation. Compared with large-scale models having from hundreds of millions to several billion parameters, it is a small model with power consumption kept from one-hundredth to one-thousandth that of LLMs. The SSM can be mounted on individual devices and edge servers to achieve autonomous control through interaction between those devices.
Application of SSM is expected to bring about a future in which the social infrastructure will be managed in an even more dynamic manner. Specifically, logistics will become more efficient and optimal through interaction between traffic signals and trucks to adjust traffic and logistics demands. In addition, mutual communication between street lamps and automobiles will simultaneously prevent traffic accidents and reduce energy consumption. In short, operating conditions such as power demands, logistics demands, and traffic demands will be analyzed in real time to make adjustments to the infrastructure automatically in a form that adapts to the environment (Figure 5).
Source: Hitachi Research Institute
Figure 5: Concept of a dynamic social infrastructure
Generate:Biomedicines in the United States is engaged in pioneering activities in the biomedical field. It has constructed a generative model for protein sequences called “Chroma” trained on hundred of millions of units of protein structures based on medical papers and open data. Chroma is attracting attention in particular for the following two functions. First, it has the ability to rapidly generate protein sequences effective in medical treatments including those nonexistent in nature. Second, it can reproduce generated sequences in 3D and verify their effectiveness and safety through more than 300 types of tests.
The application of this GenAI model specializing in drug discovery hypotheses is expected to enable personalized drug discovery in the future. It will facilitate genome analysis and microbial cultivation that combines pharmaceutical data and personal genetic information. This GenAI model will also advance the development of therapeutic drugs customized to an individual’s physical characteristics and medical history. This will lead to more effective treatments with minimal side effects and to drug discovery that contributes to the eradication of cancer and longer lives through anti-aging medicines (Figure 6).
Source: Hitachi Research Institute
Figure 6 Concept of personalized drug discovery
Archetype AI in the United States is attracting attention for its technology that supports on-site management and supervision in manufacturing processes and other areas. This startup provides a multimodal model that supports on-site operation decisions by collecting a massive amount of data such as images and sounds from on-site meters, gauges, cameras, and other devices. This multmodal model is also analyzing that data in an integrated manner. At modern manufacturing sites, it is common to perform planning optimization, image recognition, sound recognition, and other operations separately. However, the technology developed by Archetype AI combines multiple sources of information to make human-like interpretations. This technology supports even unskilled on-site managers in making decisions, similar to a veteran worker.
Widespread application of this multimodal model is expected to drive the further evolution of human and AI collaboration. In the railway field, for example, the multimodal model can integrate sensor data like pressure, temperature, and vibration in addition to image and sound data obtained from railway signals, the roadbed, and rolling stock. In this way, it will be possible to reproduce the know-how of veteran workers in tasks such as operation management, driving, maintenance, and parts procurement and support decision-making and new-employee training (Figure 7). The technology of Archetype AI can therefore be expected to not only improve business efficiency but also help raise productivity across an entire industry.
Source: Hitachi Research Institute
Figure 7: Concept of human and AI collaborative operation
In Sections 1 and 2, we focused on the bright side of GenAI given its ability to contribute to the development of society and industry. The use of GenAI, however, also has a dark side in terms of risks involving safety, transparency, and ethics. In this section, we will explore policy-driven governance. It is a framework that should be considered in development work. This framework helps address issues related to safety, transparency, and ethical standards. Its purpose is to ensure that everyone feels comfortable using GenAI technologies. Risks and concerns making up the dark side of GenAI can be divided into two parts (Figure 8). The first is technical risks related to safety and transparency such as the leakage of personal information when using GenAI, the occurrence of bias, and the generation of misinformation. To deal effectively with these problems, it will be necessary to introduce a governance system for guaranteeing transparency, a development framework that emphasizes safety, and specific rules and
regulations. Another dark side of GenAI is the psychological anxiety that it can create in people who will have to adapt to change and maybe even suffer job losses due to its penetration in society. Technological progress has the possibility of creating new social problems due to its effect on specific industries and occupations, so dealing with these problems requires ethical and philosophical research that keeps such future social changes in mind.
Source: Hitachi Research Institute
Figure 8: Risks in AI models and services and concerns of people
Governance and regulations with respect to GenAI are being established based on risk recognition and targets that differ between countries and regions. In the United States, a policy-oriented environment is being set up to promote technical innovation so that the technologies being developed by AI developers can evolve relatively freely. In contrast, the EU is enacting strict regulations such as the General Data Protection Regulation (GDPR) that emphasizes the protection of personal privacy and data safety. These standards are also being applied to AI technology.
Japan, meanwhile, is constructing original regulations and rules while referring to the approaches taken by the United States and EU. In China, the development of AI technology is moving forward in a state-led manner and dustinctive regulations are being established with the aim of ensuring national security and social order.
Amid the existence of such different approaches to regulations, the World Economic Forum (WEF) advocates a consistent development framework based on international collaboration while incorporating opinions from industry. It proposes a comprehensive framework for mitigating risks across the entire lifecycles of GenAI models to promote harmony among international regulations and guidelines. This framework can also be used as a reference when a company commoditizes its technology and can be used as guidelines for supporting the safe and ethical use of GenAI
GenAI is bringing about disruptive change in society and its rapid technological evolution is giving rise to ethical problems. Therefore, there is an urgent need for studies on methods of appropriately using GenAI and on the ethics surrounding the use of GenAI.
The Danish Design Center, Denmark’s national center for design, has formulated the Digital Ethics Compass as one means of addressing this problem (Figure 9). This Compass clarifies the ethical points involved when providing digital services and provides specific guidelines for avoiding disadvantages for users. The Digital Ethics Compass includes four basic principles and 22 specific questions and supports developers and managers of digital services in making decisions from an ethical viewpoint.
Source: Danish Design Center
Figure 9: Digital Ethics Compass for fostering ethics
Furthermore, to deal with social changes brought about by new technologies, there is also a need for an approach with a broad perspective that goes beyond short-term business results. In the United States, there are an increasing number of companies having in-house philosophers to give advice to managers. In Europe, ethics is recognized as an important element of the business process. In Japan, as well, ethical and philosophical approaches to technical issues are progressing as in the establishment of philosophy research institutes by companies.
In the above way, it is indispensable in the environment surrounding GenAI to conduct studies from an ethical perspective as technologies evolve to support the sustainable development of society. Deepening the research of technology and social changes based on ethics and philosophy should enable to appropriately deal with the changes brought on by new technologies and build an even better future.
In this paper, we used a future mandala to present how GenAI might bring about changes in the future from the viewpoints of environmental neutrality and economic growth, well-being, and collaboration between humans and GenAI. In this regard, we drew up three future scenarios: (1) Environmental neutrality and economic growth: Autonomous interoperability of digital and energy infrastructure, (2) Well-being in a super-aging society: From physical health to mental health, and (3) Business-model innovation via collaboration between humans and GenAI: Augmenting creativity by collaborating with GenAI. In each of these scenarios, the future image is one of GenAI assisting in the development and prosperity of mankind, but to achieve such a prosperous future, it will be necessary to bring together the wisdom of the world in dealing with technological and social issues. To extract the latent value of GenAI to the maximum, it is imperative that ethical problems and social issues are by appropriately addressed.
“Hitachi Souken” Journal received contributions from experts as a supplement to the research described above by the Hitachi Research Institute. These include contributions on technical innovations by GenAI from Professor Hitoshi Matsubara of Kyoto Tachibana University, social and human changes from Partner Tsuyoshi Nagayama and Principal Takuya Kagata of Dentsu Consulting, ethical aspects from Hideaki Koizumi, Emeritus Fellow of Hitachi, Ltd., and AI governance from Cathy Li, Head of AI, Data and Metaverse of the World Economic Forum. Also included in this issue is an interview with Dinesh Wadhawan, Head of Corporate Venturing Office, North America of Hitachi America, Ltd. at the front lines of AI startups.
At Hitachi Research Institute, we acknowledge the significance of technology and society mutually supporting sustainable development. In the field of GenAI, we will persist in conducting research and engaging in discussions that consider the remarkable technological advancements, social transformations, as well as governance and ethical perspectives.Through these efforts, we aim to promote the comprehensive use of GenAI by society and industry, and at the same time, to contribute to measures for dealing effectively with the risks associated with the use of GenAI.
Kenichi Shikano
Senior Manager, 3rd Research Department, Hitachi Research Institute
Kenichi Shikano is engaged in the use of IoT data usage and creation of new businesses in the fields of finance and manufacturing. He joined Hitachi in 2007. He has been in his current position since 2023 after working in macroeconomics research and related fields. Recent research themes include Digital Transformation (DX) and Green Transformation (GX) in the manufacturing and healthcare fields.
Masayasu Watanabe
Senior Researcher, 3rd Research Department, Technology Strategy Group, Hitachi Research Institute
Masayasu Watanabe took up his current position after working at Boston Consulting Group and Mitsubishi UFJ Research and Consulting Co., Ltd. His recent themes include digital platforms and digital engineering.