Digital Trust: A Key for Sustainability in the New Normal

Fujitsu Laboratories Advanced Technology Symposium 2020 report

Fujitsu Laboratories has hosted FLATS in the Silicon Valley since 2017. Over the past three years, FLATS has discussed topical themes that are important to society and highlighted latest technology developments including quantum computing and reliability of AI, as well as the balance between privacy and innovation. In addition to keynotes and panel sessions, the symposium included a technology showcase that highlighted the latest research outcomes and Fujitsu Laboratories role in furthering Silicon Valley as a center of technological innovation.

This year, given the unusual environment due to the COVID-19 pandemic, FLATS was held in an online format, and the symposium emphasized technology topics relevant to COVID-19.

Hirotaka Hara, CEO and Representative Director of FLL, set the tone for the new normal of uncertainty in his first keynote. He emphasized that resiliency was the key to a sustainable recovery from the pandemic and discussed the importance of the theme of "digital trust" and its relevance today. Dr. Hara said that trust is not just about system reliability but also about supporting change in all areas: healthcare, commerce, public health, privacy, business, sustainability, and work style both in physical and cyberspaces.

For example, Covid-19 has accelerated cashless payments including "contactless payment", wherein a credit card is not required, based on advanced biometric technology. Such new systems would not be available without trust in the entire process and the vendor.

He also cited Rachel Botsman, a social innovator and author of "Who Can You Trust?", as saying that trust will now be decentralized rather than top-down and innovations in digital technology make this possible. From new payment mechanisms to secure inter-connected supply chains, our hyper-connected world is deploying AI applications at a rapid clip however the key to their successful adoption will be the level of accountability, trust and explainability they provide so people can rely on the findings and accept the judgement provided by such systems.

FLATS focused on three major technology areas including privacy, AI and systems and software with key notes and panel sessions and corresponding themes. The three themes were "Striking a Balance for Digital Trust Between Data Privacy and Societal Prosperity", "Accelerating Towards a Trusted New Normal with AI" and "Forging a Foundation for Digital Trust with Novel ICT Systems".

Daniel Weitzner, Founding Director of the MIT Internet Policy Research Initiative, gave the second keynote titled "Two Pathways to Digital Trust: Logical Proof or Statistical Risk Management". In his talk he highlighted the fact that there were two approaches to providing digital trust. The first and traditional approach utilizes encryption and is based on mathematical or logical proofs. The second approach, currently used in credit card and ATM systems is based on statistical risk management. In this approach, although losses due to theft and malfunctions are not small, there is a mechanism to compensate for them and manage risk as a whole which could be better suited for other practical applications in the future.

Weitzner explained the importance of privacy for providing digital trust with contact tracing technology for COVID-19. In Taiwan, Singapore, South Korea, China and other countries, they succeeded in controlling the infection by using location information. However, even in the case of such applications which have immense public benefit, there may be resistance to adoption, especially in democratic countries, if there is a perception that it may cause a violation of privacy.

In March, MIT and other organizations began developing a system in which smartphones carried by people close to each other transmit Bluetooth communication signals called "Chirp" instead of location information, and then when infection is confirmed, the information is sent to the person in close contact. Mathematical proof is used to ensure that no location or personal information is exposed during collection of contact tracing data therefore the app has seen wide adoption including by Apple and Google, and used in 30 countries and 20 states in the U.S. This is a decentralized approach to contact tracing, rather than a centralized approach, and offers a way to protect privacy from health agencies.

He also offered a new perspective on compensation for intrusions into corporate information systems where in risk cannot be completely eliminated but can be managed by quantifying risk-reward. It's similar to what Benjamin Franklin created for fire insurance, where members who receive compensation after a fire must meet the structural conditions of the home that the union specifies when they join the insurance. It is also similar to the current structure of automobile safety regulations, made up of automakers, auto insurers and the National Highway Traffic Safety Administration (NHTSA).

He indicated that it would be desirable to have a digital security version of this risk-reward system for current online systems, but the problem is that companies don't disclose the frequency of security breaches and the amount of damage, so it's statistically difficult to quantify such an offering.

He is currently leading an MIT project to assess the frequency and risk of breaches within the systems of multiple companies in a way that does not compromise corporate privacy. This is an approach to secure one trust as “safety of business” and to establish another trust as “corporate credit”.

In the panel discussion in Session 1, panelists from NTT Research, Stanford University, and Boston University discussed the panel session topic from the viewpoint of blockchain technologies and contact tracing technology.

There were presentations on concepts such as decentralized trust through blockchain technologies (decentralization and multipolarization) and trust to processes in contact tracing, as well as reports of a lack of trust that makes it difficult to deploy digital technologies in healthcare settings.

As AI becomes popular in society, we have to be keenly be aware of its dangers as well as its benefits. Is there any solution to this? Stuart Russell, an AI researcher and professor at the University of California at Berkeley, argued in the third keynote that we really need to go back to the beginnings and the fundamental definition of AI.

AI technologies have been developed since 1930’s, primarily optimizing processes to achieve goals set by human beings. However, it is becoming less clear how well this standard AI approach ultimately serves humans. The reason, Russell said, is that people cannot specify objectives completely and correctly, so a machine may pursue its own goals, which could contradict what humans really want. In other words, it may turn out that the better the AI system is, worse will be the outcome for humans.

As an example, he pointed to clicks on social networking sites (SNS). With social media, we have learning algorithms that are designed to maximize click through. In order to achieve this goal, the conventional wisdom dictates that AI should display what users are interested in. But in reality, in order to achieve clickable results, AI tunes itself to make user behavior as predictable as possible.

Russell said the new model of AI should not be fixated on such short term goals but should be programmed with the ultimate objective of benefiting humans by continuously learning about human preferences. In this model, AI would be required to not interfere with human behavior and be permission based which would result in better the AI, better the outcome for humans.

This new AI model is expected to learn about the diversity of human preferences, as it is important to incorporate a wide range of knowledge into such models including ethics, economics, political theory, cognitive psychology, and brain science. To accomplish this, he concluded, it will be necessary to rebuild the foundations of AI in all areas, including search, planning, and learning. He described what AI should ultimately be, especially under the current circumstances where AI is rapidly spreading across society.

Later in the panel, panelists from Lawrence Berkeley National Laboratory, Stanford University, and Ginger presented examples of AI use in science, education, and health care. AI is now being used in materials science to test the suitability of materials for applications. By analyzing the microstructures of the material, you can predict how well it will withstand a particular application.

As for AI in education, there is a need for AI that senses what's happening in the classroom. Rather than just reacting to keyboard strokes, button presses or screen tasks, AI needs to provide teachers useful feedback about how students are engaging and offer ideas about what they want to be done differently to enhance learning.

Ginger, a startup, has developed an application software designed to address supply and demand imbalances in mental health and combines real and virtual care. AI supports diagnosis based on the tone of the user's voice and what the user is saying, by automatically displaying response options for the coach delivering care as well as evaluating the quality of the responses.

In summary the panel session highlighted that AI is expected to move to the next level and see adoption in next generation applications across a range of areas including science, education, and health care.

The last keynote was by Satoshi Matsuoka, Director of RIKEN Center for Computational Science on the theme of “Fugaku: The First Exascale & General Purpose Supercomputer Enabling Innovation”.

Dr. Matsuoka said that the name "Fugaku" which means Mt. Fuji, was intended not only to aim for a very high peak in performance, but also provide a very broad base of applicability of the system by enabling a much more general purpose capability.

To achieve this challenging goal, he described a 10-year national project that brought together RIKEN, Fujitsu, and other HPC communities to develop the system. Fugaku has won four gold medals in the latest performance rankings for key benchmarks and is already in advanced use ahead of its launch in 2021.

In addition to explaining the structure of the processor and other technical aspects, he also showed its applications. Fugaku will revamp car designs and manufacturing as well as address “Society 5.0”, a comprehensive initiative led by Japan’s Government covering healthcare, drug discovery, energy, climate change and disaster prevention. Fugaku also achieves real-time simulation running in parallel with design thereby verifying functionality and design suitability in ways that humans cannot.

Dr. Matsuoka said that in Japan, simulation videos generated by Fugaku of droplet transmission of COVID-19 and the effects of wearing a mask have been frequently broadcast in the media, leading to public awareness of the pandemic.

Following the keynote, the panel session focused on the potential of new ICT systems. PeptiDream, a drug discovery company, has succeeded in shortening the search for candidate compounds by using Digital Annealer, architecture that quickly solves combinatorial optimization problems, developed by Fujitsu.

In addition, "Project ECHO" promoted by the Stanford University School of Medicine, uses ICT to resolve disparities in health care. This project has 400 hubs around the world where healthcare professionals meet to connect with people in the community seeking medical care and advice. It is an attempt to spread limited medical resources and knowledge by utilizing non-one-on-one telemedicine.

As each panelist provided their own expertise based on their professional experience it became apparent that innovation of ICT systems and their wide range of applications and utilization will be the driving force in solving difficult problems of society in the new normal.

The need for digital transformation has been discussed over the last few years, but the COVID-19 disaster has accelerated the digitalization of all aspects of society. It has demonstrated that new ways of communicating, data privacy, and AI development are needed, along with a core underpinning of digital trust. FLATS 2020, with the breadth and depth of the sessions, brought a singular focus to these topic areas as we enter the new normal.

Author
Noriko Takiguchi
Freelance editor and journalist

Takiguchi lives in Silicon Valley. She writes extensively about technology, business, politics, international relations, design, and architecture. A special focus of her work has been trends in robotics in Silicon Valley and elsewhere in the United States. She publishes news about robotics through her own website robonews.net.
Her books include: Why is Garbage Not Sorted in Silicon Valley? (President Inc.), The Rem Koolhaas File (TOTO Publishing), and Toyo Ito, Architect Japan (Chikumashbo Ltd.). She also has translated such works as Machine of Loving Grace by John Markoff (Nikkei BP), Bringing Design to Software by Terry Winograd (Pearson Education) and Leading Matters by John L. Hennessy.
She graduated from Japan’s Sophia University, majoring in German Studies and International Relations. From 1996 to 1998, she was a Fulbright visiting scholar at Computer Science Department of Stanford University.