Creating 21st-Century Happiness with Big Data ~ Kazuo Yano, Chief Engineer, Research & Development Group, Hitachi, Ltd.

Human happiness measured by accelerometers was the strongest factor driving sales growth

──Please tell us about your own encounter with digital technology and the internet.

Yano: In 1991, during a year-long joint research project in the U.S., I witnessed researchers downloading games onto computers at distant Stanford University or accessing materials on European computers. I was amazed—how could these people do such godlike things? (laughs) That was my first encounter with the internet, and I sensed something incredible was beginning.

──So your encounter with the internet was quite early. You were also pioneering work on big data before the term even existed. What sparked your interest in that?

Yano: I had been working on semiconductors at Hitachi for a long time, but the business was phased out. I had to leave the area I was familiar with and search for a new place with colleagues. That was 12 years ago, around the time mobile phones started connecting to the internet and i-mode was beginning to spread. I thought computers would keep getting smaller, becoming wearable or attached to cars and equipment, and that's when I focused on the idea of collecting data from them.

──I see, so the starting point was wearable devices. Big data is all the rage now, but what was the situation back then?

Yano: Not at all. Not just back then, but even up until about four or five years ago, the feeling was more like, "How long are you going to keep doing something that doesn't seem profitable?" (laughs).

──Now, big data has captured the world's attention and become almost a buzzword. How do you view this situation?

Yano: I believe it's an inevitable direction. However, since we've been working on this for a very long time and have made many mistakes along the way, I sometimes feel like everyone is repeating the failures we made earlier.

──In your recently published book, you propose a fascinating theory: that human "happiness" can be measured with an accelerometer.

Yano: In an experiment where we had participants wear devices with accelerometers around their necks to collect body movement data, we found a strong correlation between the presence or absence of movement, the diversity of movement duration within a group, and the level of happiness individuals reported feeling. This diversity in movement duration is distinct from the sheer amount of movement; it's not simply that moving makes you happy. By quantifying the diversity of movement within a group, we can quantify the group's level of happiness. Conversely, wearing an accelerometer allows us to measure the degree to which a group experiences happiness.

For example, in our call center experiment, even when handling identical products, days with greater diversity in employee movement and above-average happiness levels showed significantly higher order rates compared to days without such diversity. We then investigated in more detail how to enhance this sense of happiness. We found very clearly that workplaces where employees moved around more during breaks—meaning lively chatter was happening—had higher happiness levels. Furthermore, when the overall workplace happiness was high, the entire workplace's order rate increased.

──Your book also mentions sales experiments in retail stores besides the call center. Using big data revealed that strategies long considered industry standard were largely ineffective, while unexpected areas showed significant impact. This must have been a shocking result for the advertising and marketing industries. How did you personally feel about it?

Yano: We've seen this pattern quite often in other industries too, so honestly, we're used to it. But without such discoveries, there'd be no point in investing in big data, right?

──Having worked with big data for over a decade, what are the key points to pay particular attention to when handling it?

Yano: The crucial point is that data is meaningless unless you understand its quality and value. It must be data that assists in making decisions about specific actions to take. Otherwise, even vast amounts of mere facts aren't very useful. However, data that directly connects to value is surprisingly scarce. But I believe we've broken through that barrier by being able to measure happiness. By cross-referencing this happiness value data with various factual data, we can judge the merits of specific behaviors—like whether arriving early to work is beneficial.

When value data combines with factual data, the factual data truly comes alive. Otherwise, it becomes merely data used to back up what humans want to verify – data for justification that lacks significant impact. And this kind of value data must be actively sought out; it won't come unless you make the effort. The value differs depending on the problem definition each company or situation aims to address. Value is determined by humans. However, Happiness holds special impact as a universally applicable value, independent of the company.

A turning point has arrived, shifting from a century-defining idea to another

──Still, "happiness leads to profit" is quite a groundbreaking discovery, isn't it?

Yano: I believe happiness is the ultimate, highest value. It's understood as a concept in self-help, psychology, philosophy, and religion. We've found that happy people are also highly productive. But I don't want anyone to misunderstand: people who are just taking it easy or being lazy aren't happy. Boredom isn't a good state for humans.

In other words, people cannot be happy without just the right amount of challenging tasks before them. It's somewhat natural that being happy leads to higher productivity, and the data shows this objectively and without bias. There are also various other real-world examples demonstrating that increased happiness leads to higher sales.

──Understanding this suggests significant changes in how people work, including workplace layouts and organizational design.

Yano: That's right. We can now clearly provide evidence-based answers to questions like: How many subordinates should a section chief have? Is it better for section chiefs to stay late? Should meetings be kept short?

──As big data utilization expands, will there be differences by industry?

Yano: Currently, big data utilization is relatively advanced in areas where both output and process data can be captured. Marketing is a prime example of this domain, with widespread use of linked data like daily retail sales and product display information. The call center example mentioned earlier generates massive amounts of output and process data daily, making it, in a sense, an ideal testing ground.

More challenging are areas with a time lag between process and output, such as corporate sales. In these cases, identifying the right leading indicators becomes crucial. Agriculture and healthcare also fall into this category. Even more difficult is the realm of knowledge work, where the output itself is often indefinable. However, our research has identified happiness as a potential substitute metric in this domain, suggesting big data utilization will likely expand here too.

──If such developments progress, they could lead to significant shifts in values, couldn't they?

Yano: That's precisely what we want to change—the world. We're now at a turning point, shifting from 20th-century thinking to 21st-century thinking. The 20th century strongly favored seeking one universal rule applicable everywhere. But reality involves far more context-dependent situations. The 20th century was about standardizing know-how and scaling it up: measuring processes and outputs like "how many units per hour" on manufacturing lines to boost productivity.

However, in knowledge work and services, both output and process are hard to see. From the latter half of the 20th century until now, we tried to address this by sharing manufacturing manuals and know-how. But in actual workplaces, I believe people have been devising their own solutions rather than applying uniform methods. That's precisely why big data is effective in knowledge work and services. On the other hand, big data isn't really necessary for mass production. The results we obtained from our store experiments are only valid for that specific store under those specific conditions. They change when new variables, like seasonal shifts, are introduced. Even within the same store, conditions differ, and they vary even more between different stores or industries.

When dealing with such complex, situation-dependent factors, big data and artificial intelligence are overwhelmingly strong against such fluctuations. They enable finding the optimal solution tailored to each specific situation and location. This means continuously collecting data and constantly evaluating it using AI. I've recently started using the term "total AI-ization," and I believe all business will undergo this total AI-ization.

──If that progresses, the nature of marketing and advertising communication will likely change too.

Yano: Absolutely. I believe it will fundamentally transform productivity in advertising, marketing, and related fields.

 

Can big data provide more human-like answers than humans?

──Will the world of big data change as various types of wearable devices emerge in the future?

Yano: Absolutely. Being able to capture every moment of a person's 24/7, 365-day life means capturing their entire behavior pattern, which is an enormous impact.

──IoT (Internet of Things) and robots are also gaining attention. Will robots also have a major impact on big data?

Yano: Robots are essentially concentrated artificial intelligence. The mechanical parts like motors and batteries haven't advanced that much; what's advancing is the computer and AI technology inside them, and that will continue. I believe the value of robots lies in how AI, while being used within computer systems, also interacts with the real world.

──While big data includes numerical values from sensors, internet posts are also utilized as big data. Going forward, it seems the use of data will expand beyond text to include images and videos. As data types diversify and volumes increase, will the capabilities of big data fundamentally change?

Yano: It will change. Until now, computer programs simply operated according to the hypotheses set by their designers. Even if circumstances changed, the program wouldn't adapt unless updated. However, as computers gain the ability to learn from diverse data, they themselves will adapt and change in response to changing situations. Software emerged 80 years ago, but now we're witnessing a new leap: software itself is beginning to collect data, alter its own logic, and evolve.

──Big data is deeply connected to artificial intelligence. How do you view its current trajectory?

Yano: Artificial intelligence is at a historic turning point. Since the Turing machine appeared in the 1930s, software has matured over the past 80 years. But that era involved humans writing logic and then deducing data and actions from it. The ability to learn inductively from real-world data represents a 180-degree shift. AI went through a winter period, but I believe it's finally becoming truly substantial.

However, I believe the common discussion in AI debates about the Singularity—the idea that AI will surpass humans in the near future—is fundamentally flawed. The argument goes that computers will eventually encompass all human knowledge, becoming all-knowing and perpetually outperforming humans. Yet this assumes somewhere that the cost of acquiring information becomes zero. We handle projects starting from the data acquisition phase, so we understand well that this cost becoming zero is impossible.

For instance, capturing all the nuances and nonverbal meanings in human conversation is difficult, and even if possible, it wouldn't be cost-effective. Humans live, using their five senses to sense their surroundings, and that's free. There's no way a computer could acquire the information humans gather about their environment at a lower cost. No matter how advanced artificial intelligence becomes, it cannot transcend the laws of physics.

Ultimately, while some information known to humans overlaps with what AI knows, there will always remain significant information known only to humans. Therefore, we should focus our energy on how humans can effectively utilize AI as a valuable tool, and on how to combine big data with the unique insights and knowledge only humans possess. AI is meant to be a kind of partner or sherpa for humans; it's about figuring out how to climb higher together.

The image of big data providing mechanical, human-ignoring answers while humans possess warm hearts is mistaken. Take the earlier call center example: the experiment was conducted to boost sales. Yet the solution proposed by big data and AI was to create an organization where employees enjoy lively chatter during breaks and everyone is happy – a far more human-centered answer than humans themselves could have come up with (laughs). Data also has a positive side: it encompasses various things without being bound by prejudice.

Not interruption technology, but happiness technology.

──Regarding expectations for big data, there's the aspect of predicting the future. There are real-world examples like predicting election results or influenza outbreaks. Do you think it will expand into even more areas going forward?

Yano: Those examples aren't about predicting the future; they're about making optimal decisions now while considering the future. The future involves countless variables, and gathering all the relevant data and information is cost-prohibitive. For instance, predicting that a specific incident will occur at a specific location is impossible. While influenza outbreak predictions do exist, they're less about prediction and more about extracting accurate leading indicators from the data. Such applications are already commonplace and will undoubtedly advance further.

──Attempts to create artistic works like music using big data and artificial intelligence are already underway. What do you foresee for the future?

Yano: For instance, if you play synthesized music to about ten people, measure their happiness levels, and then provide feedback to increase that happiness, you could likely create music that way. We might see more initiatives like this emerge in the future.

──Privacy concerns are often discussed regarding data collection. What are your thoughts on this?

Yano: It's an important issue, but personally, I feel it's a bit premature. That is, focusing solely on the concerns before figuring out how to utilize the data is unbalanced. We need to also consider the positive aspects in a balanced way. As technology evolves, human values naturally change. For example, 20 years ago, when companies started giving employees cell phones, there was widespread criticism about feeling tied down.

However, as they rapidly became widespread, the positive aspects emerged. When weighing the pros and cons, the positives outweighed the negatives, leading to a societal consensus in favor of the technology. I believe we'll see similar situations arise in various areas going forward.

──Earlier, you mentioned that conventional wisdom based on human experience was largely incorrect in those store and call center experiments. What will happen to human experience and intuition as technology advances further?

Yano: Regarding the store experiments, it's humans who set the challenge of increasing average customer spend rather than just attracting more customers. Humans decide such challenges and the information fed to the AI. Humans also determine the next steps based on the results and bear responsibility for those outcomes. I think the approach will be to amplify the power of that experience and intuition using tools.

──So technology has the power to correct biases humans have held, but it doesn't take everything away; there will still be things humans need to do.

Yano: Whenever new technology has changed the times, the portfolio of jobs needed has naturally shifted before and after. This isn't unique to AI; it's an inevitable consequence of innovation. It's unavoidable, and we need to focus on how to achieve a soft landing. Of course, there are areas where AI can perform better than humans in certain professions, so that's something we should consider. On the other hand, new occupations will certainly emerge. For example, new skills and professions focused on leveraging AI for business development will inevitably arise.

Consider Plato, the ancient Greek philosopher. In his dialogue "Phaedrus," he criticized the invention of writing as harmful. Replace "writing" with computers, the internet, or artificial intelligence, and his critique holds true today. Humans are like this: we repeatedly rely on tools, causing certain abilities to decline, only for new abilities to emerge in their place.

──Recently, there's also debate about whether humans should distance themselves more from digital and the internet. Indeed, the volume of information flowing through the internet has already surpassed what humans can reasonably read, right?

Yano: Well, the solution is simple: just don't read it (laughs). It's about how to allocate human time effectively. However, there is a real problem: many current tools are designed specifically to capture people's attention. This unknowingly robs people of various abilities. In one experiment, they compared productivity between groups with email notifications enabled versus disabled. The group with notifications had lower productivity. So the problem is these interruptions to attention and concentration. I think even children might be feeling this impact lately.

──I see. So it's not that digital or the internet are inherently bad, but rather the structure of these tools that constantly interrupt people is problematic.

Yano: One form of human happiness is being able to immerse oneself, to become engrossed and passionate about something. Being constantly interrupted and cut off is detrimental. It would be healthier for happiness-enhancing technology to become widespread and gain greater influence than interruption technology. That's what I aim to achieve.