AI ECONOMIC SALVATION OR SOCIETAL PERIL?

Transcription

1 AI ECONOMIC SALVATION OR SOCIETAL PERIL? Should we welcome or fear the arrival of artificial intelligence? Mark Purdy Managing Director, Economic Research Accenture Research Image: Kai Stachowiak Bath Royal Literary and Scientific Institution 23 April 2018

2 INTRODUCTION SPEAKER Mark Purdy, Managing Director - Economic Research Accenture Research Copyright 2018 Accenture. All rights reserved. 2

3 FOR MANY OF US, OUR FIRST INTRODUCTION TO THE IDEA OF ARTIFICIAL INTELLIGENCE CAME VIA THE MOVIES AI in service to the society or AI posing threats to humanity? TARS, the space robot from Interstellar (2014) v.s. Robotics fighting back from I, Robot (2002) Crime forecasting from Minority Report (2002) Baymax, the healthcare robot from Big Hero 6 (2014) A supercomputer deeming mankind a threat, Singularity (2017) Machines taking control of Earth, the Matrix (1999) Copyright 2018 Accenture. All rights reserved. 3

4 HUMAN-LEVEL AI IS STILL A LONG WAY OFF What AI can and can t do ARTIFICIAL NARROW INTELLIGENCE Beat Go world champions ARTIFICIAL GENERAL INTELLIGENCE Understand abstract concepts Read facial expressions Explain why Write music Be creative like children Diagnose mental disorders Tell right from wrong Comfort earthquake survivors Have emotions MORAVEC S PARADOX Copyright 2018 Accenture. All rights reserved. 4

5 WHAT IS ARTIFICIAL INTELLIGENCE Sense AI TECHNOLOGIES Computer Vision Audio Processing ILLUSTRATIVE SOLUTIONS Identity Analytics Virtual Agents Natural Language Processing Cognitive Robotics Comprehend Knowledge Representation Speech Analytics Machine Learning Recommendation Systems Expert Systems Data Visualization Act Copyright 2018 Accenture. All rights reserved. 5

6 AI HAS A LONG HISTORY OF SUMMERS AND WINTERS The 1st boom Reasoning as research Natural language WABOT-1-world s first full-scale intelligent humanoid robot The 2nd boom The rise of expert systems The knowledge revolution Neural networks make a comeback 2000-present Deep learning, big data and artificial general intelligence The birth of AI Cybernetics - thinking machines The Turning test Dartmouth conference Symbolic reasoning The 1st AI winter Limited computer power Limited database capacity Limited networking capabilities Real-world problems are complicated The 2nd AI winter A sudden collapse of specialized AI hardware vendors Nouvelle AI and embodied reason Copyright 2018 Accenture. All rights reserved. 6

7 Number of Priority Patents RECENTLY IT HAS BEEN UNDERGOING ANOTHER BOOM Both AI startup investment and patent publication on AI have accelerated since 2010 AI Startup investment growth is outpacing that of the overall startup market ( ) AI investments ($) AI deals (#) Filed Priority Patents Artificial Intelligence named filings and related technology filings AI keywords (atleast one AI technology directly named in patent filing) AI smartsearch (AI related tech filings) 9X 6X 10,000 Global AI vs overall startup investment markets ($) AI startup investment Overall market startup investment 54% vs. 16% 8,000 6,000 4,000 2, % By 2016, 11% of all global startup deals were AI-oriented, up from 2% in c* Priority Year *Note: For year 2015, only 6 months data is published and full year data is assumed at a factor of two Copyright 2018 Accenture. All rights reserved. 7

8 f 2018f 2019f 2020f BECAUSE OF THREE DRIVING FORCES Rapid increases in computational power Huge fall in cost of storing data near zero marginal cost Explosion in data that is digitized One dollar s worth of computer power, Hard drive cost per gigabyte, Global installed base of IoT units (number of units, billions) Billions of computations per second (log scale) USD (log scale) 1,000, Business Consumer 100, , , Source: The Brookings Institution Source: mkomo.com Source: Gartner, Forecast: Internet of Things - Endpoints and Associated Services, Worldwide, 2016 Copyright 2018 Accenture. All rights reserved. 8

9 THE RAPID ADVANCE OF AI HAS LED TO HEATED DEBATES AROUND ITS POTENTIAL THREATS AND OPPORTUNITIES AI in the news: The potential economic benefits and risks How people feel about AI? Public discussion has exploded since 2009 In general, AI has had consistently more optimistic than pessimistic coverage over time Articles on AI Optimistic articles Pessimistic articles Source: Long-term trends in the public perception of AI, Fast & Horvitz, 2016, Association for the Advancement of Artificial Intelligence Copyright 2018 Accenture. All rights reserved. 9

10 COULD AI BE AN ECONOMIC SALVATION? 10

11 GLOBAL ECONOMIC GROWTH IS WEAKENING GDP growth (real, % y/y) Developed economies Emerging economies World Source: Oxford Economics Copyright 2018 Accenture. All rights reserved. 11

12 LOOKING BACK, GDP GROWTH HAS STEADILY SLOWED IN MANY LARGE ECONOMIES SINCE THE 1980S Real GDP growth (%, annual average over the period) s 1990s 2000s 2010s Japan United States United Kingdom France Italy Germany NB: Data points across the dashed lines indicate the average for the six countries. Source: Oxford Economics Copyright 2018 Accenture. All rights reserved

13 THIS IS BECAUSE THE TRADITIONAL FACTORS OF PRODUCTION ARE UNDER PRESSURE Capital efficiency: The marginal capital efficiency rate has steadily dropped over a 50-year period. Labour: As populations age and birth rates slow, fewer people are available to pick up the slack in the workforce Marginal capital efficiency (%, 6-year moving average) Japan United States Germany United Kingdom Working age population (%, annual average growth over the period) Copyright 2018 Accenture. All rights reserved United States United Kingdom France Italy Germany Japan

14 AND PRODUCTIVITY GROWTH IS TRENDING DOWNWARD A key measure of how well an economy uses its existing capital and people is total factor productivity (TFP). Data show a weakening of TFP, especially in the past 10 years. Total Factor Productivity (%, annual average over the period) United States 3.0 Japan 3.0 Germany France Copyright 2018 Accenture. All rights reserved United Kingdom Italy Source: The Conference Board, Total Economic Database

15 SO ARE WE EXPERIENCING THE END OF GROWTH AND PROSPERITY? Copyright 2018 Accenture. All rights reserved. 15

16 AI CAN BE A POWERFUL REMEDY FOR SLOWING GROWTH AI has the potential to overcome the physical limitations of capital and labour and open up new sources of value and growth Traditional Growth Model Δ Capital ΔLabour TFP Growth Adapted Growth Model ΔCapital Δ Labour TFP Δ AI NB:Δ indicates the change in that factor. Source: Accenture analysis Copyright 2018 Accenture. All rights reserved

17 AI CAN BOOST NATIONAL ECONOMIC GROWTH VIA THREE CHANNELS Intelligent Automation Augmentation Innovation Diffusion Copyright 2018 Accenture. All rights reserved. 17

18 AI PROMISES TO BOOST NATIONAL ECONOMIC GROWTH AI has the potential to double annual growth rates in the countries we analyzed in terms of gross value added (a close approximation of GDP). Real gross value added (GVA) (%, growth) Source: Accenture and Frontier Economics Baseline AI steady state Copyright 2018 Accenture. All rights reserved

19 CONCERNS AND FEARS AROUND AI ARE ON THE RISE 19

20 AI IS ON THE CUSP OF TRANSFORMING EVERY ASPECT OF HOW WE DO OUR JOBS AND LIVE OUR LIVES HOW WE WORK HOW WE LIVE Labour market disruption Future of jobs Erosion of individual identity Data privacy Bias and fairness Growing inequality Post-work society Copyright 2018 Accenture. All rights reserved. 20

21 CONCERNS THAT TECHNOLOGY WILL DESTROY OUR JOBS ARE NOT NEW David Ricardo, The Principles of Political Economy and Taxation the substitution of machinery for human labour, is often very injurious to the interests of the class of labourers. Karl Marx, Capital this effect of machinery represented to be a compensation to the working class, is on the contrary a most frightful scourge. John M. Keynes, Economic Possibilities for our Grandchildren We are being afflicted with a new disease... namely, technological unemployment. This means unemployment due to our discovery of means of economising the use of labour outrunning the pace at which we can find new uses for labour. Kurt Vonnegut, Player Piano A novel that depicts a world dominated by a supercomputer and run completely by machines Paul Samuelson, Ricardo was Right! Frey & Osborne, The Future of Employment: How Susceptible are Jobs to Computerisation? McAfee & Brynjolfsson, Race against the Machine Copyright 2018 Accenture. All rights reserved. 21

22 HISTORICALLY, TECHNOLOGY HAS AFFECTED LABOUR MARKET TRANSITIONS IN FOUR DIFFERENT WAYS Substitution Effect Industrial robots took over manual tasks such as welding, painting and assembling Complementary Effect Computers improved efficiency and created new jobs such as computer-aided designers Income Effect Full-time earnings for British workers more than doubled in the seventy years after 1780 Price Effect Transportation innovations and development sharply pushed down the cost of travelling Copyright 2018 Accenture. All rights reserved. 22

23 Lead Empathize Create Judge Train Explain Sustain Amplify Interact Embody Transact Iterate Predict Adapt NEW EMPLOYEE ROLES AND NOVEL TYPES OF WORKING RELATIONSHIPS ARE REQUIRED The missing middle: new jobs, roles, and skills will be required as organizations use AI to augment and transform the way they work HHuman-only activity Humans complement machines AI gives humans superpower Human and machine hybrid activities M Machine-only activity Source: Human + Machine: Reimagining Work in the Age of AI, Daugherty & Wilson, Harvard Business Review Press, 2018 Copyright 2018 Accenture. All rights reserved. 23

24 AI BECOMES DEEPLY INGRAINED IN OUR EVERYDAY LIVES, YET REAL CHALLENGES REMAIN REGARDING PRIVACY, TRUST AND EXISTING BIASES Consumers want to be more engaged in managing their data 87% of consumers believe it is important to be able to review and control their personal data online. Slaves to the algorithms? Reinforcing existing stereotypes, social and cultural segregation and exclusion Personalized recommendation leading to a filter bubble? 3/4 however, do not find it easy to manage that data. Source: Accenture Dynamic Consumers, 2017 Copyright 2018 Accenture. All rights reserved. 24

25 BUT AI COULD BE THE PIVOTAL TECHNOLOGY IN ADDRESSING THE GRAND CHALLENGES OF HUMANITY Inclusive education and personalized learning Knewton, an AI-enabled adaptive learning platform leverages machine learning to personalize content to each user to improve academic outcomes. Empowering the disabled Seeing AI, a free mobile application designed to support people with visual impairments by narrating the world around them. Fight against world hunger Urban traffic management Stanford University has instituted an AI program that uses machine learning to predict crop yield. The goal is to provide planters with information allowing them to make more thorough planting and food reservation decisions. The AI traffic system in Pittsburgh- Surtrac which can detect traffic and allow traffic lights to adapt to realtime conditions has helped cut travel time by 25% Health management and disease diagnosis DreaMed Diabetes, an Israel-based startup, developed an algorithm which continuously monitors patients glucose levels and defines precisely when and how to adjust insulin levels. Fight against crime The Murder Accountability Project (MAP) has developed an online serial-detection algorithm. Established in 2015, this serial-detection system is designed to identify areas with significantly low clearance rates through the recognition of the homicidal patterns of individual serial killers. Copyright 2018 Accenture. All rights reserved. 25

26 How to realize AI s promise while controlling the peril It all depends on how we manage the transition to an era of AI 26

27 FORGING A DIGITAL NEW DEAL THAT ALLIES ECONOMIC PROSPERITY WITH SOCIAL PROGRESS A digital 'New Deal - Three priorities Big Ambition Big Inclusion Big Skills Agenda Imagine the future Address grand challenges Catalytic investment Regulatory innovation Responsible AI Missing Middle skills Regional & sectoral adjustment Grassroots support Reskilling for Jobs of the Future Social and Technical skills Transition paths for affected workers Life-long education system Copyright Accenture. Accenture. All rights All rights reserved. reserved. 27

28 ANNEX 28

29 READY FOR HUMAN-MACHINE COLLABORATION? Employers underestimate the willingness of employees to acquire the relevant skills 62% 67% I believe intelligent technologies will create opportunities for my work. It will be important/very important to learn new skills to work with intelligent technologies in the next 3 to 5 years. ONLY 3% of executives say they intend to significantly increase investment in training and reskilling programs in the next three years. Sample: More than 1,200 CXOs, more than 14,000 workers, 12 industries Source: Accenture, Reworking the revolution, 2017 Copyright Accenture. Accenture. All rights All rights reserved. reserved. 29

30 AI WILL BE BOTH A DISRUPTING AND ENABLING FORCE IN THE WORKPLACE Over the years, technology has made some jobs redundant, but also led to the creation of new jobs and occupations: Change of share of employment in the OECD area, (percentage points) But the impact of AI is more than automation; It s more about workplace augmentation Free us up from 3 D s Dull, Dirty, Dangerous High-skill occupations Middle-skill occupations Low-skill occupations 7.6 Legal analysis 1.9 Make us more productive and innovative -9.5 Source: OECD, OECD Employment Outlook 2017 Virtual assistants AI surgical robots Copyright Accenture. Accenture. All rights All rights reserved. reserved. 30

31 AI RISKS EXACERBATING INCOME INEQUALITY AND POLARIZATION OF SOCIETIES ACROSS THE WORLD Technological progress has contributed the most to widening income inequality across the globe from mid- 1980s to mid 2000s (average annual percent change) Change in inequality And a new digital divide is looming on the horizon Internet penetration (%), Contribution of trade and financial globalization Contribution of technology 20 Contribution of other High Income Countries Low Income Countries Middle Income Countries World Source: IMF Source: OECD Copyright 2018 Accenture. All rights reserved. 31

32 LOOKING AHEAD, AI WILL POSE NEW CHALLENGES TO BOTH REGULATORS AND BUSINESSES Limits of Taylorism Robot taxes Transition paths to new jobs Universal basic incomes Learn with the machine Digital inclusiveness Copyright 2018 Accenture. All rights reserved. 32