transforming industry: the role of artificial intelligence

41
TRANSFORMING INDUSTRY: THE ROLE OF ARTIFICIAL INTELLIGENCE Dirk Pilat Deputy Director Directorate for Science, Technology and Innovation [email protected] 7th Concordi Conference: Innovation for Industrial Transformation Seville, 25 September 2019

Upload: others

Post on 06-Feb-2022

2 views

Category:

Documents


0 download

TRANSCRIPT

TRANSFORMING INDUSTRY: THE

ROLE OF ARTIFICIAL INTELLIGENCE

Dirk Pilat

Deputy Director

Directorate for Science, Technology and Innovation

[email protected]

7th Concordi Conference: Innovation for Industrial Transformation Seville, 25 September 2019

Outline

1. AI – context, applications and some trends

2. Transforming industry – some key issues for policy and analysis

3. OECD’s work on AI

4. Launch of JRC-OECD report: World Corporate Top R&D Investors

1. CONTEXT, APPLICATIONS AND SOME TRENDS

What is Artificial Intelligence?

4

• No agreed definition: AI consists of simulating certain learning processes of human intelligence, to learn from it and replicate it. AI rests on two key elements: data and algorithms (and computing power).

• OECD definition of an “AI System”: An AI system is a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments.

• Today’s Artificial Intelligence is “narrow” or “applied” and designed to accomplish a specific problem-solving or reasoning task. Even the most advanced AI systems available today, such as IBM’s Watson or Google’s AlphaGo, are still “narrow”.

• Applied AI can be contrasted to a (hypothetical) Artificial General Intelligence (AGI), in which autonomous machines would become capable of general intelligent action, like a human being, including generalising and abstracting learning across different cognitive functions.

Today, the science underlying AI continues to advance …

5

Source: OECD calculations based on Scopus Custom Data, Elsev ier, Version 1.2018 and 2018 Scimago

Journal Rank from the Scopus journal title list (accessed March 2018), January 2019. Statlink:

https://doi.org/10.1787/888933928445

Trends in scientific publishing related to artificial intelligence, 2006-16 Index of publication counts, 2006=100

100

120

140

160

180

200

220

240

260

2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

Index 2006 = 100

Total scientific publications Scient ific publications related to AI

0

20

40

60

80

100

2006 2016

%CHN EU28 USA Other

Share of world's AI publications

.. with EU countries making an important contribution

6

Source: OECD calculations based on Scopus Custom Data, Elsev ier, Version 1.2018 and 2018 Scimago

Journal Rank from the Scopus journal title list (accessed March 2018), January 2019.

0

5

10

15

20

25

30

35

% Top 10 economies

2016 2006

0

0.5

1

1.5

2

2.5

%Other economies

Top-cited scientific publications related to AI, 2006 and 2016 Economies with the largest number of AI-related documents among the 10% most

cited publications, fractional counts

Patenting is increasing rapidly

7

Source: OECD, STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats, January 2019,

Statlink: https://doi.org/10.1787/888933928331

Technology developments in artificial intelligence, 1990-2016 Index 1990=1 based on the number of IP5 patent families

0

2

4

6

8

10

12

Index 1990 = 1

AI-related paten ts All patents

0

20

40

60

80

100

Identification of AI-relatedpatents, 2012-16

%

IPC codes onlyIPC codes and keywordsKey words only

Transport: autonomous vehicles with potential cost, safety and

environmental benefits

Science: accelerate discovery, facilitate reproducibility and lower

experimentation costs.

Health: detect health conditions, deliver preventative services, discover

treatments, facilitate clinical research, and optimise health systems.

Criminal justice: predictive policing, assessing recidivism risk, and

predicting court procedure outcomes.

Security: detect anomalies and threats in digital security and surveillance.

Agriculture: agricultural robots, crop and soil monitoring and impact

prediction.

Financial services: detect fraud, assess credit worthiness, automate

trading, etc.

Marketing and advertising: target and personalise content, advertising

and products, demand forecasting, translation, etc

And, of course, economic analysis.

AI has a wide range of applications, …

… as emerges from patenting data, …

9

Top fields of application of AI-related technologies, 2012-16

Share of application fields in AI-related patents, IP5 patent families

Source: OECD, STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats, January 2019;

Statlink: https://doi.org/10.1787/888933928350

0

1

2

3

4

%Other technology fields

0

10

20

30

40

% Top 5 fields65

… with AI combining with many other technologies

10

Top technologies combined with AI, by field of application, 2012-16

Share of technology (IPC) classes in AI-related patents, IP5 patent families

Source: OECD, STI Micro-data Lab: Intellectual Property Database, http://oe.cd/ipstats, January 2019;

Statlink: https://doi.org/10.1787/888933928369

0

1

2

3

4% Other technologies

0

2

4

6

8

10

12

14% Top 5 technologies

Computer technologies Audiovisual technologies IT methods Telecommunications Medical technology Digital communication Control

Private investment in AI start-ups is growing

11

Total estimated investments in AI start-ups (USD billion), 2011-2018 (Jan-Jun)

By start-up location

Source: OECD estimates based on Crunchbase (July 2018), www.crunchbase.com, Note: Estimates for 2018 may be conservative, as they do not account for a likely lag in reporting.

… with the UK accounting for the bulk of start-up investments in the EU …

12

Investments in AI start-ups based in the EU (USD billion), 2011-mid-2018

Percentage of total amount invested in EU-based start-ups over period

Source: OECD estimates based on Crunchbase (July 2018), www.crunchbase.com,

.., and investments spanning a range of industries

13

Top sectors of AI Start-ups from 2011 to 2017

Source: OECD estimates based on Crunchbase (April 2018), www.crunchbase.com,

2. TRANSFORMING INDUSTRY - KEY QUESTIONS FOR POLICY AND

ANALYSIS

OECD’s Going Digital Framework can help structure thinking about AI

1. Who has access? – Mainly to data (but also algorithms and computing)

• AI relies on, and leverages data in fundamentally new ways

• Network and scale effects

How to enhance access to data

• Public research data

• Curated and accurate data

• SME access

• Public interest and global challenges (e.g. a Global Data Commons)?

• Markets for data

For example, machine learning requires large amounts of training data

17

https://medium.freecodecamp.org/chihuahua-or-muffin-my-search-for-the-best-computer-vision-api-cbda4d6b425d

2/3. Use and Innovation: can AI help strengthen productivity growth?

Probably, but:

– The diffusion of AI has only just started.

– AI may contribute to the productivity divergence between leaders & laggards

– Its impacts will depend not just on AI technologies, but also on organisational change, skills and new processes.

– It will require structural change to enable the growth of AI business models, including flexible regulations.

– All of this may take time to have an impact – experience suggests that productivity may initially slow down.

0.00

0.05

0.10

0.15

0.20

0.25

Digital intensity Sh. High-skilled KIS

Low High

The rate of catch-up in productivity is lower in digital and skill intensive industries

Source: Berlingieri et al., 2019, Last but not least: laggard firms, technology diffusion and its structural and policy determinants

Source: OECD (2019), Measuring the Digital Transformation, based on OECD, ICT Access and Usage by Businesses Database, http://oe.cd/bus, December 2018.

Enterprises performing big data analysis, by size, 2018 As a percentage of enterprises in each employment size class

Large gaps across and within countries in technology use, e.g. big data …

4. How will AI affect jobs and skills?

AI and automation typically only affect certain tasks; most jobs involve many tasks. Highest risk in routine jobs with low skill and education requirement BUT many jobs will be affected

Source: OECD, Science, Technology and Innovation Scoreboard 2017, based on OECD, Education Database, September 2017.

0

2

4

6

8

10

% Women Men

Tertiary graduates in information and communication technologies, by gender, 2015

As a percentage of all tertiary graduates

And how can countries foster AI talent?

• Privacy: Profiling, monitoring,

automated decision-making,

algorithmic bias

• Safety and responsibility: What do

concepts such as “product”, “safety”,

“defect”, “damage” mean for self-

learning and autonomous systems?

What about cybersecurity risks that

could impact safety (e.g. if hackers

take control of car)?

• Transparency: Understanding and

explaining how systems operate,

which factors influence result, level of

certainty. Detecting bias. Being able to

challenge results.

5/6. Society and Trust: How can we make

AI trustworthy?

7. Market openness and competition:

how does AI affect competition? Value of M&A deals per Year by Target Firm Industry

Source: Zephyr M&A database

Value of M&As in Digital Sectors – Normalised (2005 = 100)

3. OECD’S WORK ON AI

• Agreement on Council Recommendation on Artificial Intelligence [C/MIN(2019)3]] adopted by 42 countries (36 members + 6 others) at the OECD Ministerial Council Meeting of 22-23 May 2019, and also the basis for G20 AI Principles, as agreed by Digital Economy Ministers on 8 June. Work now underway on implementation of the principles.

• Measurement and analytical work – report on AI in Society – 11 June 2019

• Establishment of an AI “Observatory”, with several pillars of work, to be launched in Q4 of 2019

OECD work on Artificial Intelligence

AI Policies & Initiatives

AI PolicyObservatory

Jobs, skills, data, etc.Health, transport, etc.

National AI policiesInitiatives

OECD metricsPartner metrics

AI Principles

OECD AI PrinciplesPractical guidance

AI Public Policy Topics

AI Metrics & Measurement

Governments that have adhered to the OECD or OECD-based G20 AI principles

OECD members Adherents G20 principles, based on OECD

4. JRC-OECD REPORT ON WORLD CORPORATE TOP R&D INVESTORS

WORLD CORPORATE TOP R&D INVESTORS:

SHAPING THE FUTURE OF TECHNOLOGIES AND OF AI

2019

• Relies on EU-JRC Scoreboard 2017 data

on world’s top 2 000 R&D investors (2016),

linked to IP assets and scientific publications (NEW!)

• 2019 edition:

Broad overview of the innovative activities

of top R&D investors

Changes between 2012 and 2016 samples

Special focus on developments in Artificial Intelligence

2019 report: Shaping the future of technologies

and of AI

Share of economies in the top 2000 R&D companies

Location of world’s top R&D investors,

Headquarters, 2016

Source: JRC-OECD, ca lculations based on EU Industrial R&D Investment Scoreboard (2017), May 2019.

Share of economies in total number of subsidiaries

Location of world’s top R&D investors,

Subsidiaries, 2016

Source: JRC-OECD, ca lculations based on EU Industrial R&D Investment Scoreboard (2017), May 2019.

Share of economies in the top 2000 R&D companies

Cumulative percentage shares within the top 2000 R&D companies, 2014-16

R&D investment, publications, IP bundle

Distribution

0

10

20

30

40

50

60

70

80

90

100

0 250 500 750 1 000 1 250 1 500 1 750 2 000

%

R&D ranking of companies

R&D expenditures Patents Scientific publications Trademarks

Note: Data relate to companies in the top 2 000 corporate R&D sample, ranked by R&D investment in 2016. The IP bundle refers to the number of patents and trademarks

fi led in 2014-16, and owned by the top R&D companies, and the number of scientific articles are those published by authors affiliated in the top R&D companies during the same time-period, using fractional counts. Source: JRC-OECD, COR&DIP© database v.2., 2019.

Share in total IP5 patents, trademarks & publications, resp.

Patents, trademarks and publications

of world’s top 2000 R&D investors, 2014-16

Source: JRC-OECD, COR&DIP© database v.2., 2019.

Top R&D investors Other

Patents Trademarks Publications

Leading companies:

publications, patents, and trademarks, 2014-16

Top 50 patenting Top 50 publishing Top 50 trademarking

Source: JRC-OECD, COR&DIP© database v.2., 2019.

Changes among the top R&D investors

2012 versus 2016

Top 20 companies

Note: Bold orange lines indicate an increase in ranking of more than 10 positions.

Source: JRC-OECD, ca lculations based on EU Industrial R&D Investment Scoreboard (2013 and 2017), May 2019.

New top R&D investors, by sector & HQ, 2016

AI developments by top R&D investors, 2014-16

Contribution of world top R&D investors

to Artificial Intelligence

Source: JRC-OECD, COR&DIP© database v.2., 2019.

Top 50 patenting

0

25

50

75

100% Top R&D investors

Top 50 AI patenting companies own 54% of AI patents

Share of AI in patents, trademarks and publications,

top R&D investors and other actors, 2014-16

Contribution of world top R&D investors

to Artificial Intelligence

Source: JRC-OECD, COR&DIP© database v.2., 2019.

0.0

0.5

1.0

1.5

2.0

2.5

Publications Patents Trademarks

% Top R&D investors Other

Leading sectors in AI:

patents, TMs and publications

Top 5 sectors in AI, word top R&D investors, 2014-16

Source: JRC-OECD, COR&DIP© database v.2., 2019.

0

10

20

30

40

50

%Patents

0

10

20

30

40

50

%Publications

0

10

20

30

40

50

%Trademarks

• Raw data

• Free download

• 3rd version of COR&DIP©

JRC-OECD COR&DIP© database v.2, 2019

World Top 2000 Corporate

R&D Investors

Financial2013-16

R&D, Sales, Employees, etc.

Trademark Portfolio2014-16

EUIPO, USPTO

Trademark Classes

NICE Class

Scientific Publications

Counts by ASJC fields

Patent Classes

IPC & WIPO technologies

Patent Portfolio2014-16

IP5 families

Further information and

data access: http://oe.cd/ipstats http://iri.jrc.ec.europa.eu/other-reports.html

THANK YOU!

OECD team

Hélène DERNIS Shohei NAKAZATO

Brigitte VAN BEUZEKOM Mariagrazia SQUICCIARINI

JRC team

Petros GKOTSIS Nicola GRASSANO Antonio VEZZANI

Thank you

41 41

Contact: [email protected]

OECD Going Digital website: http://oe.cd/goingdigital

OECD work on AI: http://www.oecd.ai