Kazuo Yano of Hitachi and David Rejeski of the Environmental Law Institute discuss the applications and challenges of general-purpose artificial intelligence. | Andrea Korte/AAAS
Artificial intelligence can boost business productivity and profitability — as well as find ways to foster human happiness, according to the director of Hitachi’s artificial intelligence laboratory.
Interest in AI traditionally has centered on purpose-built technology crafted for a particular reason, such as Google’s AI program that last year beat a champion player at the Chinese strategy game Go, said Kazuo Yano during a 14 December presentation at AAAS headquarters.
The address was part of the Hitachi lecture series, which has brought speakers to AAAS for nearly a decade to examine a wide range of issues related to science and society.
Yano, who serves as Hitachi’s chief corporate scientist, noted that AI increasingly is being used to address the needs of business. To cope with changing variables like customer behavior and marketplace position, businesses now can take advantage of the flexibility offered by “general-purpose AI,” which can be added to existing systems, he said.
For business owners trying to increase store sales “every day is new,” Yano added.
For such business applications, humans define the parameters and the desired outcome, Yano said, and the AI component then establishes hypotheses and takes actions based on live data streams, quickly adapting to changing factors like supply and demand and learns along the way. Unlike conventional software, general-purpose AI relies on varying logic that adapts to specific needs.
“This endless, tireless pursuit of the outcome of learning gets results,” Yano said.
Yet general-purpose AI’s very agility – which allows it to be used in any number of circumstances – raises questions about how it might be received and regulated.
As the use of AI by business becomes more common, it is never too soon to proactively consider and seek to address the policy implications the technology raises such as concerns about privacy and human job displacement, said David Rejeski, a director at the Environmental Law Institute, who served as a discussant responding to Yano’s lecture.
Yano noted, however, that people are the ones who determine and establish the objective AI is intended to target, said Yano, who has also written extensively on how AI can quantify and improve human happiness.
Yano, who has used activity trackers to record his own movements for nearly a decade, has found that self-reported happiness correlates very strongly with diverse physical motion – that is, people who start and stop movements, however small, at irregular intervals report greater happiness than those who spend long, regular periods of time either moving or at rest.
If we can quantify happiness, Yano said, we can also use general-purpose AI to help us better achieve it.
Hitachi’s own general-purpose AI system, known as H, has been used to assist in a number of different situations in a diverse range of industries like transportation, e-commerce and stock trading, Yano said.
“It can solve a very wide range of problems,” Yano said.
For example, H was added to a retail store’s point-of-sale system in order to improve sales numbers. The store incorporated the professional advice of retail specialists, who suggested putting advertisements at key locations throughout the store, Yano said.
Analyzing the data and coming up with its own hypotheses, the computer generated an alternative solution: employees should stay in one place longer. The proposal was put into practice and the AI-derived solution improved sales by 15 percent compared to the retail specialists’ solution that led to flat sales, Yano said.
This type of general-purpose AI will permit what Yano called a “total digital cycle,” as AI begins to take on not just the automation of tasks but the learning required to improve how tasks are carried out.