The Rise of AI Robotics

Japan’s path forward as physical objects gain intelligence

15 January 2025

Japanese: 1 October 2024

Takuya Mori
Center for Advanced Technology
Hirohiko Nakamura
Center for Advanced Technology

AI robotics is a burgeoning field defined by robots equipped with AI for advanced cognitive functions. Riding the momentum created by the boom in generative AI, big tech leads the pack of companies piling into AI robotics. Meanwhile, the traditional robotics industry is also accelerating AI-based development, further fueling the competition to dominate the field. What lies beyond is a world of intelligence permeating all objects, transcending the conventional robots of industry and customer service. In this way, robots will become ingrained in daily life, otherwise forgotten to be extraordinary entities. In light of where the latest technology is going, Japan needs to face down a few challenges to adapt to a world running on AI robotics.

Competition for dominance in AI Robotics

As interest in AI booms worldwide and people notice the huge sums big tech and startups are pouring into AI development, the area of greatest competition is shifting from AI software to AI robotics, the integration of AI tech into robots. The spread of robots with sophisticated intelligence could spur innovation and help resolve societal issues in ways not previously possible, especially in medical and longterm care and in performing complicated tasks in high up and hazardous places.

Innovating all sorts of applications built on large language models (LLMs) trained on massive datasets, big tech and their friends have established the current AI business model premised on AIs to substitute for humans in performing knowledge-intensive tasks. They believe that they can extend their successes further in AI robotics. Specifically, they are building foundation models trained on diverse categories of data accumulated using robots that interface with the real world. And they envision that, by fine-tuning their AIs, they can develop general-purpose robots capable of performing tasks following instructions given in natural language, which is inherently ambiguous.

Of course, the road to general-purpose robots is not going to be without bumps and potholes, and the required investment is astronomical. But the tech giants and others like them, accepting the challenge, are braving the risks and aiming for hegemony in the real world beyond the internet, in technologies to replace humans in production theaters as well.

Different industries bring different approaches

As AI industry players burst into robot development, robot industry players, with years of technical innovation already under their belts, are increasingly incorporating AI into their work. The combination is accelerating the pace of AI robotics development.

The word robot made its debut in 1920, illustrating just how far back the history of robots themselves goes. Since the first industrial robots appeared in the 1960s, research and development has flourished in Europe, Japan, and the US, giving rise to the academic field of robotics. Each of the field’s major elemental technologies—intelligence, control, motors, and sensors—have seen steady strides forward. More recently though, with innovation getting more incremental, robot research is showing signs of maturity. It is also experiencing an upturn, greatly accelerated by robotics players’ building on the burst of progress in AI tech.

The AI industry and the robot industry each have their own approach to their products: Building on its success with generative AI, the AI side tends to be oriented toward AI robotics for more general-purpose applications. Over in the robot industry, the tendency is to focus on hardware applications specialized for specific tasks or industry sectors while also making use of foundation models (Figure 1). The competition resulting from these different approaches is boosting the commercialization of more-attractive AI robots.
Figure 1: Typical approaches of the AI and AI robotics industries
Typical approaches of the AI and AI robotics industries
Mitsubishi Research Institute, Inc.

Tech defined by its role, robot a term of the past

When people mention robots, they’re usually talking about industrial robots; namely, robots that perform specific, automated tasks in cordoned-off areas of factories. Less often, they’ll be talking about service robots, the kind that assist people in some way, though these—like their market size and pervasiveness in society—are generally secondary. But the rapid emergence of AI robotics and robots’ concomitant democratization is changing that. It is bringing about a society in which non-specialists readily use robots with little thought about it. That is because AI robots go beyond being the mere servants like those that replace people or assist them in performing their jobs more efficiently. AI robots are people-friendly partners that can respond empathetically to delicate shifts in users’ state of mind or physical condition. This democratization of robots will of course spark growth in markets for service robots, but down the road it is also likely to bring about a shift to incorporating intelligence-endowed objects into just about everything in people’s real-world surroundings (Figure 2).
Figure 2: From AI robotics to intelligent objects
From AI robotics to intelligent objects
Mitsubishi Research Institute, Inc.
Until recently only specialists came into direct contact with robots, but the technology is changing. Robots are gaining the ability to empathize and becoming a familiar part of everyone’s environment. Not so far away is a world where robots blend into daily life, even going unnoticed as some sort of special presence. The evolution of AI robotics is moving humanity into an age where even the very word robot is a thing of the past.

Japan’s challenges: Investing capacity and customer expectations

Smooth progress in AI robotics technology alone is not going to bring about this rosy future, particularly in Japan.

A number of bottlenecks can be posited. Foremost is Japan’s capacity for investing in AI robotics. Unlike conventional AI, AI robotics entails hardware, which in turn entails limits on how far costs can be reduced. This is especially so during the initial stages of uptake, making sufficient investing capacity indispensable for reaping the benefits of AI robotics.

Domestic sectors like agriculture and healthcare, where using robots holds much promise for addressing labor shortages, are good examples of why. There has been little progress in the pooling of capital in them, which means investment in AI robotics is likely to get pushed down the priority list. In fields like these, what matters going forward is how much and how rapidly those engaging in them can evolve into entities capable of making the needed investments.

Another bottleneck is customer quality expectations. As with generative AI, key to progress in AI robotics is using diverse datasets to train the AIs. Put conversely, during the early stages of diffusion there will be limits to the precision of AI robots. The technology will need to be improved, which can only be done incrementally with data accumulated by using it in the real world. That means customers will have to tolerate imprecision for a while. What matters most is whether Japanese people and corporations, notoriously unwavering for quality, can be innovative customers, actively utilizing AI robotic products and services rather than allowing themselves to be put off by quality shortcomings.

A receptive domestic market would be a boon for Japanese enterprises to compete on the world stage in AI robotics. Getting past these two bottlenecks—investing capacity and customer expectations—will be pivotal for building that kind of AI robotics-philous environment. Japan needs more than supply-side industrial policies aimed at academia and the robotics industry to enhance its industrial competitiveness in the AI robotics field. The government should move to develop and implement demand-side policies targeting users of AI robotics, such as pooling capital and engineering change in quality expectations.

Author profile

Author

Takuya Mori

Center for Advanced Technology

Takuya Mori has been involved in creating new businesses oriented toward super-aged society and conducting research on policy for science, technology, and innovation, industry–academia partnership, and university reform. His recent work primarily targets the healthcare field, spanning support for national R&D projects, academic-startup business development, and active-senior career development. He has a doctorate in education.

Author

Hirohiko Nakamura

Center for Advanced Technology

Hirohiko Nakamura joined Mitsubishi Research Institute in 1989 and has been involved in technology-oriented research and consulting, particularly materials and electronic devices. He has covered a wide range of fields including nanotechnology, materials, neutron science, and the utilization of outer space. Since 2020, he has been researching virtual technologies and AI robotics, among other topics, at the Center for Advanced Technology. He has a Doctor of Science degree.