AI has now entered into a new phase known as generative AI. Beginning with the arrival of large language models (LLMs) such as ChatGPT, the scope of AI’s use is continually growing to include things like image and video generative AI and speech generative AI, and this development holds the potential to expand our own human intellectual abilities. NRI has grouped these expanded intellectual capacities into six categories: “Predict”, “Distinctify”, “Customize”, “Communicate”, “Model”, and “Create”.
 
We are already seeing examples around the world of how these six abilities are being expanded. In the near future, we can likely expect to see these same kinds of examples occur in Japan as well. In particular, with Japan facing a deepening workforce shortage amid population decline, there will need to be discussions about leveraging AI to maintain the quality of services, and AI will have to be utilized in a variety of everyday-life situations.

If AI were to permeate our daily lives, how might our everyday-life situations change?
When you want to dine out, for example, an AI agent would consider your preferences and plans, the weather that day, and predicted congestion based on customer flow data and then decide on a restaurant. When you arrive, an AI “staff” member that knows how many patrons are in your party shows you to the optimal table. When taking your orders, it then checks each person’s preferences regarding ingredients, flavors, cooking methods, and so on, and then comes up with a recipe based on the available ingredients. Even with this one example of dining out, we can envision how AI may be used in any situation from choosing a restaurant to serving food. This concept has already begun to germinate, and through the combination of multiple functions, such AI “staff” could possibly even handle a whole series of tasks.
 
The use of AI also holds promise when it comes to solving the labor shortage in medical settings.
For example, when a patient comes to the hospital, a hospital check-in notification will get delivered via a facial recognition gate, and his or her completed medical questionnaire is then sent to a physician. During the exam, an AI doctor inquires about the patient’s symptoms while referencing his or her medical history, and then transcribes that information into the patient’s clinical records. When prescribing medicine, the AI will use the patient’s chart and genetic information to predict any potential future illnesses, and then formulate the optimal medicine for that person. We are also already seeing such cases begin to emerge, and although at present these are limited to partial applications of AI, we may expect that eventually, AI will handle the entire process, thereby eliminating the labor shortage in the medical field.

While we are seeing AI coming into greater use, the recognition and use of generative AI still remain at low levels, and according to the results of NRI’s questionnaire survey, 48% of respondents said they “don’t understand the details”. Although AI has the potential to be utilized in various everyday life settings, when we asked the survey takers how much they were willing to accept AI in different everyday life scenarios (whether they would want to use AI or would approve of the use of AI), the reply given by some 20-30% of respondents for each given scenario was that they did not want the use of AI at all, which revealed that there are certain psychological hurdles to be overcome. Some of the areas where respondents were comparatively more receptive to the use of AI included “household chores”, “transit (driving cars)”, and “government (counter services at city hall, etc.)”; a conceivable factor for such high levels of receptiveness is that there are already examples of products and services that make use of AI in these scenarios, which makes it easier for people to imagine AI being utilized there.
 
Meanwhile, it is also apparent that the lack of understanding about AI has led to these low levels of acceptance, and thus among those who had never heard of generative AI, more than half responded that they “do not want to use AI at all”. Spreading understanding of AI will likely be the key to addressing this psychological hurdle.
That said, there are also some psychological hurdles that cannot be eliminated merely with a greater understanding of AI, such as concerns over AI’s precision, uncertainties regarding where responsibility lies when there are problems, and people’s appreciation of the value of having a human being help them. For instance, with regard to everyday-life situations involving people’s well-being, livelihoods, or assets (e.g., healthcare/nursing, government, finance), many respondents said that “I don’t want to use AI because it’s hard to know where the responsibility lies when problems occur”, whereas when it comes to everyday-life situations premised on manned services (lodging/dining out/schooling, etc.), most respondents said that “I don’t want to use AI because I think having someone there to help is part of the service”. When it comes to understanding why people are reluctant for AI to be used in a given everyday-life situation, we must first grasp the background behind why their receptiveness to AI is so low.
 
In our survey, we also inquired about public perceptions of AI. When we checked people’s level of trust regarding the AI agent “AI Great Men”, which has recreated images of great figures from history, 39% of respondents said they could fully or at least partly trust those images. Furthermore, when asked whether they “think AI can be controlled by human beings” or whether they “believe AI has a will”, there were variations in the responses. This suggests that people’s perceptions regarding AI still vary considerably.

Based on the questionnaire results, NRI has defined the fields in which AI will see the most widespread use going forward, plotting the data on the two axes of “interpersonal communication” and “physical or mental load”. The fields that most readily lend themselves to AI use are those where the tasks involved are most easily fixed and generalized, and that require relatively little interpersonal communication. In particular, those activities or fields involving high physical or mental loads such as household chores, automated driving, and diagnostic imaging can be said to be opportunities for AI adoption. Conversely, in fields like nursing or childcare with both their high need for interpersonal communication and high physical/mental loads, the psychological hurdles concerning the use of AI are also high, and these are also fields where people appreciate the value of having human beings provide those services, which makes it harder for AI to be readily utilized.
 
Given the history of how new technologies spread in the past (mobile phones, etc.), there are three barriers standing in the way of AI’s further widespread use, namely “accessibility”, “transformation of values”, and “getting the surrounding environment to keep up”. It would seem that the first barrier, “accessibility”, has already been overcome.


As for “transformation of values”, in the fields mentioned above where AI is more apt to be utilized, we must pay attention to how AI stands to change everyday-life situations which it had previously been assumed would always be handled by human beings. We may also anticipate that as the use of AI spreads further, we will see a transition from partial optimization to overall optimization. Just as the manufacturing industry has become more specialized, we can expect that there will be similar divisions of labor among different AIs, leading to performance improvements.
 
Lastly, with AI just beginning to take off, in order to dispel the psychological hurdles revealed in our survey, we hope to see an appropriate level of human involvement to provide a sense of security as new “killer contents” come out that stand to transform society’s values.