Society 5.

0 and Artificial Intelligence with a Human Face

Society 5.0 and Artificial Intelligence with a
Human Face
December 3rd, 2019

Indian Council for Research on International Economic Relations (ICRIER)

Core 6A, 4th Floor, India Habitat Centre
Lodhi Road, New Delhi -110 003
 Objectives of the study
 An assessment of AI ecosystems across G20 countries

 Application of AI to sustainable development goals (SDGs)

 Challenges of using AI towards achievement of SDGs

 Policy recommendations and the way forward

 Introduction : Society 5.0

Society 5.0
Society 1.0 Artificial
Society 2.0 Society 3.0 Intelligence to
Hunter- Society 4.0
Agrarian Industrial expand human
Gatherer Digital Society capabilities
Society Society and address
Society
social
challenges

Social problems identified by Japan

•Reduction of Greenhouse gas (GHG) emissions

•Increase production and reduce loss of foodstuffs Society 5.0 was proposed in the
5th Science and Technology
• Migration of costs associated with ageing society Basic Plan as a future society
that Japan should aspire towards
•Promotion of sustainable industrialization

•Redistribution of wealth and correction of regional inequality

 Artificial Intelligence (AI)
AI has characteristics of a general purpose
technology (GPT) that has the ability to
predict and automate pre-defined sets of tasks
using and in some case generating data over a
particular task (Taddy 2018)

• PWC estimates 14% additional growth

($15.7 trillion) of global GDP over existing
trends by 2030 due to AI

• Accenture predicts doubling of developed

country GVAs due to AI

• AI to add US$957 billion to India’s

economy in 2035 : Accenture
 AI in the G20
• Discussions on the digital economy began with Chinese presidency in 2016. The outcomes
focused on the need to intensify efforts to measure the digital economy and its consequences
to create supportive policy environments
• The German presidency (2017) also focused on the roadmap for digitisation as well as the
potential of the digital economy to support the 2030 Agenda for Sustainable Development
• The Argentina presidency (2018) furthered the discussion on strengthening the digital agenda
for development
• At the G20 Ministerial Meeting on Trade and Digital Economy, held on 8–9 June 2019 in Japan,
the G20 Trade and Digital Economy Ministers outlined G20's commitment to a human-centred
approach to AI, guided by a series of G20 AI Principles.
• These Principles have a non-binding nature and are drawn from the OECD Recommendation
on AI (adopted in May 2019).
• Several G20 countries have independently proposed the creation of a council or a committee
or a task force for standards and regulations on the ethical use and development of AI.
• The Osaka Track at the G20 Summit 2019 attempted to legitimise informal plurilateral
negotiations on digital trade including rules on free flow of data (South Africa, India and
Indonesia decided to opt out)
 Country led strategy for AI adoption and Implementation
Role of Government

• Among the countries that have a policy on AI, many have prescriptive policies with specific targets. 10 countries have also
committed investments for the development and use of AI.

• G20 countries including China, India, Japan, European Union, France, Germany and United Kingdom have policies explicitly
aimed at promoting social and inclusive growth.

• Indonesia has not yet published a dedicated artificial intelligence strategy. However, leading in the Asia-Pacific market in
adoption of AI with 65 percent businesses responding to a survey stating, they had planned or already implemented AI
solutions.

• Saudi Arabia became the first country to grant citizenship to a robot in October 2017
 Role of the Private Sector and Academia
Number of AI Patents Granted by Country Distribution of AI startups across some of the G20 countries in
8000 (2000-2016) 2018
Number of AI Patnets granted by

7000 1600
6000 1393
1400
5000
4000 1200

Number of AI startups
country

3000
1000
2000
1000 800
0
600
383
400 245
200 42 45 55 82 106 109 113 131
19 22 26 27
0

Source: WIPO 2019 Report

Number of paper on AI published in Academic

Journals: 1996-2018 Source: Data compiled from Asgard and Berger “Artificial
Intelligence – A Strategy for European Startups
140000
AI patent filing in academia
published in Academic Journals

120000
AxNumber of paper on AI

100000
• As per WIPO 2019, of the top 20 universities and public research
between 1996-2018

80000
organizations in the AI field, 17 are located in China and the
60000
remaining three in the Republic of Korea
40000
20000
Private Investments in AI based startups
The European…
The United…

0
Brazil
Australia

Italy

India
Canada
Russia

Mexico

France
Indonesia

Finland

Japan

China
Argentina

Turkey

South Korea
South Africa

Denmark

Germany

The United States

Saudi Arabia

• Private companies dominated investment in AI with internal

corporate funding accounting for USD 18 billion –USD 27 billion
of total AI funding. Funding from venture capital and private
equity funds amounted to approximately USD 6-9 billion, while
Source: Scimago Journal USD 2-3 billion came from mergers and acquisition.

• United States accounts for 70-80 % of the global venture capital

investment across all technologies
 AI for Good
SDG Goal Use Cases AI tools Country Impact
Goal 1: No Poverty. Sustainability and Sensor Rwanda, Nigeria, Feeding an algorithm with both night and day satellite images of
Artificial Uganda, Malawi and Rwanda, Nigeria, Uganda, Malawi and Tanzania, were able to
Intelligence Lab: Tanzania identify signifiers in the daytime pictures. Combining both image
Project by sets, this helped the computer predict poverty in these regions,
Stanford with an 81-99% accuracy as compared to the night time satellite
University images alone.
Goal 2: Zero Hunger Farmwave Image processing Australia Farmwave provides smart image processing to present growers
real-time feedback at every stage of crop maturity. By accessing
the web app, one can use the built in tools to scan and identify pest
and disease issues, or perform rapid counts for yield estimation.
The app has observed 1000+downloads
Goal 3: Good Health Google Machine Learning UK, Canada, USA, Over 100 million people are affected by diabetic retinopathy or
and Well-Being. DeepMind Health France age-related macular degeneration. These conditions can cause
sight loss unless they’re treated quickly. The results, which were
published in Nature Medicine, showed that Deep Mind AI system
could recommend patient referrals as accurately as world-leading
expert doctors for over 50 sight-threatening eye diseases.

Goal 4: Quality Liulishuo Machine Learning China Liulishuo offers a personalized and adaptive language learning
Education experience to 50 million learners in 379 cities in China and over
175 countries around the world
Goal 5: Gender Textio Natural Language USA Australian software giant Atlassian used Textio's software for its
Equality. Processing job-ad copy which resulted in 80% increase in the hiring of
women in technical roles globally over a two-year period.”
Goal 6: Clean Water EMAGIN Sensors Canada In emerging markets, Emagin has been working with utilities to
and Sanitation. support technological leapfrogging, and has successfully deployed AI
in non-digitized utilities. In established markets, Emagin has
supported progressing facilities beyond digitization by implementing
AI to fully optimize operations and save on expenditure. After
winning a series of awards locally, Emagin was recognized in the
international water sector by BlueTech and Global Water
Intelligence.
 AI for Good
SDG Goal Use Cases AI tools Country Impact
Goal 7: Affordable Ecoisme : Smart Natural Language UK and used in The startup provides an electricity bill forecast,
and Clean Energy home App Processing Australia, Asia, measures the amount of electricity used by
Africa, North and different devices, and allows one to switch their
South America utility provider which results in saving the energy
bills by 15%.
Goal 9: Industry, PARC Machine Learning Palo Alto, PARC delivered to Olay a suite of algorithms and
Innovation and California, USA user experience elements and, in September of
Infrastructure 2016, the Olay Skin Advisor platform was released.
The platform enabled accurate analysis of users’
skin, informed users of what was happening with
their skin, suggested product and regimen changes,
and provided a compelling user interaction flow.
Over 4 million site visits worldwide. Users of the
platform exhibited 2x the conversion rate and 40%
larger basket size upon checkout as compared to
regular Olay.com visitors. They also yielded 3x
lower bounce rate and 4x time spent as
engagement measures.
Goal 10: Reduced SEEING AI- Sensors 56 countries Seeing AI helps users (blind or with low vision)
Inequality Microsoft read printed text in books, restaurant menus, street
signs and handwritten notes, as well as identify
banknotes and products via their barcode.
Leveraging on-device facial-recognition technology,
the app can even describe the physical appearance
of people and predict their mood. The app has
observed 697 downloads since its launch
 AI for Good
SDG Goal Use Cases AI tools Country Impact

Goal 11: Sustainable PetaBencana.id Sensors Indonesia Since its debut in 2013 (as PetaJakarta.org), the
Cities and PetaBencana.id platform has been used by millions of
Communities resident users to make time-critical decisions about safety
and navigation during emergency flood events; it has also
been adopted by the National Emergency Management
Agency (BNPB) to monitor flood events, improve response
times, and share time-critical emergency information with
residents. Currently it supports a coverage area with over
50 million residents in Jabodetabek, Surabaya, and
Bandung,

Goal 12: Responsible OSCAR Computer vision and Canada According to a benchmarking waste study of a university’s
Consumption and Machine Learning contamination rates and with early Oscar performance in
Production waste diversion, it is approximated that by nudging users
to recycle correctly, each Oscar installed in a facility can
reduce up to 10 tons of waste from being sent to landfill
and thus stopping toxic chemicals from seeping into
surrounding water tables

Goal 13: Climate Grid4C Machine Learning USA In a few short years they have scaled to 4 continents and
Action are partnering with global utilities to analyze billions of
meter and smart devices reads to deliver millions of
predictions every day. As the #1 ranked predictive
analytics solution provider in the energy industry.

Goal 15: Life on Land Resolve Image Processing United States The TrailGuard AI technology was field-tested at the
and Deep Learning Grumeti Reserve in Tanzania, where it enabled the arrest
of thirty poachers and the seizure of over 1,300 lb. of bush
meat.
Goal 16: Peace and Quantexa Machine Learning , UK HSBC is using AI to help spot money laundering, fraud, and
Justice Strong Deep Learning and terrorist funding. AI enables the bank to screen the vast
Institutions Natural language amounts of data it holds on customers and their
Processing transactions against publicly available data, in the search
for suspicious activity.
 Challenges of using AI
Bias and Ethical Use of
Skills and Inequality Scalability
AI

• Studies find that tasks • Algorithmic bias’ wherein AI • Most AI applications to

performed by workers that programs systematically yields developmental cases are either at
are routine in nature are results that are discriminatory an experimental stage or are
most likely to be automated, restricted to a particular area or
leading to worker demography
• Achieving social justice will be
displacement and more difficult as AI cannot bear
consequently a rise in any ‘responsibility’ or • Scalability of such applications is
inequality (Acemoglu and culpability for errors or for limited by access to data and
Restrepo, 2018) unfairly excluding victims. crucially, on access to quality data.

• Automation is by definition • Novel AI programs are also • Deficiency in quality data, which is
labor substituting being heavily deployed for scarce, can be linked to deficiencies
purposes of surveillance, in institutional infrastructures that
• As a consequence the manipulating elections, engage in collection of data sets
demand for high skilled will subverting democracy, and
further exacerbate extant curtailing individual liberties. • Data protection regulations that,
inequalities within the while aiming to protect individual
workforce, and between privacy, also makes it more
employers and employees challenging for companies that
provide AI based solutions
 Downside Challenges of AI: Can AI hold us captives?

• The rapid adoption of humanoids and AI enabled home companions

has raised concerns on how machines may eventually replace human
interaction

• Studies have found that when robots fail to reciprocate human

emotions, feelings of loneliness could be exacerbated instead of
alleviated

• The level of social disconnection can also endanger health. Loneliness

has been alleged to have the same impact on life expectancy as
smoking 15 cigarettes a day, with a risk factor that rivals excessive
drinking or obesity

• Data from United States finds that the introduction of digital platforms
and smartphones, has led to an increase in the suicidal rates among the
youth aged 10 to 17 by 70 % between 2006 and 2016
 Way Forward and Scope for Collaboration
Strengthening AI Building an All Invest in Social
Building Addressing Ethical and
Research and Encompassing Sector
Capacity Adoption of AI Accountable AI
Development Data Strategy Applications
•The upgradation •Forming • Targeted •Developing a • Standards have
of skills requires cooperative policies focused generalized the ability to •Social sector
amendments in relationships on enhancing meta-data influence the applications may
the education with some of the literacy, relevant standard for development not always find
curriculum frontrunners in local language data.gov.in will and deployment immediate
•The ‘G20 AI—such as US, content and enable of AI based monetisation
Education China, Japan, UK applications, integration of systems by opportunities.
Ministers’ and Germany affordability and resources mandating There is a need
Declaration availability of including but product for continued
2018′ •Widen the scope high speed not limited to specifications focus on social
encouraged of academic digital data, tools, for sector
participation, collaboration. infrastructure is literature, etc. explainability, applications of AI
especially of vital for robustness, and
women, in STEM harnessing the •The experiment fail-safe design
fields. AI ecosystem with Data Trusts
•Internship and in the UK is
Apprenticeship worthy of
programs to consideration
prepare the
emerging labour •Open Data has
force for been advocated
industry by several G20
countries
including United
States, France,
Korea, etc.
Thank You !