In this blog post, we’ll examine the anticipated challenges and countermeasures at each stage by dividing the development of artificial intelligence into early, mid, and late phases.
The prediction that “by 2045, AI will surpass humanity…” is a claim Ray Kurzweil has made on numerous occasions. In his book ‘The Singularity Is Near’ and elsewhere, he argues that the Singularity—the moment when technology develops so explosively that it becomes impossible to predict with human reasoning, thereby surpassing humanity—will eventually arrive. As evidence that the Singularity will inevitably occur, he cites the Law of Accelerating Returns and has predicted the future based on past trends in technological development and statistical data.
If the Singularity does arrive, we will be able to manipulate matter at the molecular level, and the development of artificial brains and AI robots will become an unstoppable trend. So, should we simply stand by and watch passively? Certainly not. As the Singularity approaches, serious social and ethical issues will arise at each stage, so we need to recognize these in advance and prepare for them.
Some seek to block development altogether through legal regulations, citing the risks of artificial intelligence. However, completely halting technological progress is practically impossible. Therefore, it is crucial to formulate countermeasures tailored to each phase of the inevitable changes and, in particular, to respond quickly and decisively in the early stages.
When a new paradigm emerges, it tends to develop slowly at first as a reaction against existing practices, accelerate rapidly in the middle phase, and finally stabilize as the pace of development slows. Based on this, we can consider the advent of an AI society in terms of early, middle, and late phases. The early stage is when AI designed for specific purposes, such as self-driving cars, becomes widespread; the middle stage is when general-purpose strong AI—capable of learning and making judgments like humans—and the hardware equipped with it (e.g., Android) emerge; and the late stage is when the singularity is reached, leading to the formation of a superintelligent society featuring cloud-based brain-computer interfaces.
In the early stage, purpose-driven AI begins to replace relatively simple tasks. Attempts to commercialize self-driving cars are already well underway, and some companies are operating pilot services. Consequently, the most significant social issue is the impact on jobs held by low-skilled, low-wage workers. Looking back at past projections, an international forum warned in a previous presentation that many jobs could disappear despite the creation of new ones, and a report based on technological forecasts at the time predicted that software, robots, and smart machines would likely replace certain occupational groups in the future. It is worth noting that job losses are expected to be particularly significant in office and administrative roles involving a high volume of simple, repetitive tasks. Historically, there have been precedents where workers’ resistance and social unrest arose when society failed to respond appropriately to technological change.
However, there is no need to be entirely pessimistic. In fields related to the Fourth Industrial Revolution—such as big data, cloud computing, the Internet of Things (IoT), platforms, 3D printing, and VR/AR—there is a high probability of an explosive growth in new jobs. Therefore, a realistic approach is to first investigate which occupations are most likely to be replaced and determine the size of the workforce in those fields. Subsequently, policies should be formulated to limit further hiring in those sectors and gradually transition to automation, while providing AI-related education and technical training to unskilled workers and the next generation of talent to prepare them for redeployment into new jobs. Responding proactively in this way can help mitigate the widening wealth gap and social conflict.
In the medium term, the emergence of general-purpose strong AI could intensify polarization among nations and disputes over the benefits and harms associated with AI. For example, the development of technologies such as military combat robots will spark debates over the issue of liability. It would be extremely dangerous to begin addressing these issues only after AI has already emerged. Therefore, discussions regarding the status, rights, and responsibilities of AI must begin in advance, and concrete laws and sanctions must be established through collaboration not only among AI developers but also with sociologists, ethicists, and futurists. Furthermore, coordination efforts are needed at the international level, such as establishing consultative bodies to set certain limits on the commercialization of AI by corporations and to prepare for potential disputes. Since such international coordination and consultation take a long time, advance preparation is essential.
In the later stages—that is, once the singularity has been reached—it is more realistic for humanity to adapt quickly to AI based on the solutions gathered during the early and middle stages, rather than attempting to delay the process by suppressing AI. The major impacts of the late stage can be summarized into two main categories. One is the issue of identity confusion, as seen in “You 2”; when an AI nearly identical to a human is created, identity confusion may arise between the individual and their replicated counterpart. This problem can be mitigated through a deep understanding of and education regarding both AI and humans. The other is the emergence of cloud services that materialize matter at the molecular level and connect to the human brain. In a superintelligent society, the level of reflection on ethical issues will improve alongside technological advancement; therefore, issues related to cloud-based brain connectivity must be reexamined based on the issues resolved in the early and middle stages.
In summary, while the exact year of the Singularity may be controversial, the transition to a superintelligent society appears inevitable. We have divided this transition into early, mid, and late stages, identified the representative problems of each stage, and proposed simple solutions. These solutions need to be further refined and supplemented, and they do not guarantee a complete resolution. However, rather than simply being optimistic about the future or standing idly by, efforts are needed to minimize these problems. Experts from various fields—including not only AI developers but also sociologists, ethicists, and futurists—must come together to establish countermeasures in advance of technological development.
