Artificial intelligence (AI) can transform the productivity and GDP potential of the global economy. Strategic investment in different types of AI technology is needed to make that happen.
Labour productivity improvements will drive initial GDP gains as firms seek to "augment" the productivity of their labour force with AI technologies and to automate some tasks and roles.
Our research also shows that 45% of total economic gains by 2030 will come from product enhancements, stimulating consumer demand. This is because AI will drive greater product variety, with increased personalisation, attractiveness and affordability over time.
The greatest economic gains from AI will be in China (26% boost to GDP in 2030) and North America (14.5% boost), equivalent to a total of $10.7 trillion and accounting for almost 70% of the global economic impact.




1. Artificial intelligence (AI) is a source of both huge excitement and apprehension. What are the real opportunities and threats for your business? Drawing on a detailed analysis of the business impact of AI, we identify the most valuable commercial opening in your market and how to take advantage of them. Sizing the prize What’s the real value of AI for your business and how can you capitalise? +14% PwC research shows global GDP could be up to 14% higher in 2030 as a result of AI – the equivalent of an additional $15.7 trillion – making it the biggest commercial opportunity in today’s fast changing economy. +26% The greatest gains from AI are likely to be in China (boost of up to 26% GDP in 2030) and North America (potential 14% boost). The biggest sector gains will be in retail, financial services and healthcare as AI increases productivity, product quality and consumption.

2.Defining AI In our broad definition, AI is a collective term for computer systems that can sense their environment, think, learn, and take action in response to what they’re sensing and their objectives. Forms of AI in use today include digital Human in the No human in assistants, chatbots and machine learning loop the loop amongst others. Automated intelligence: Automation of manual/cognitive and routine/non- Hardwired Assisted Automation routine tasks. /specific Intelligence Automation of Assisted intelligence: Helping people to systems AI systems that manual and assist humans in cognitive tasks that perform tasks faster and better. making decisions are either routine Augmented intelligence: Helping people to or taking actions. or non-routine. make better decisions. Hard-wired This does not Autonomous intelligence: Automating systems that do involve new ways decision making processes without human not learn from of doing things intervention. their interactions. – it automates existing tasks. As humans and machines collaborate more closely, and AI innovations come out of the research lab and into the mainstream, the Adaptive Augmented Autonomous transformational possibilities are staggering. systems Intelligence Intelligence AI systems that AI systems augment human that can adapt decision making to different and continuously situations and can learn from their act autonomously interactions with without human humans and the assistance. environment. For a full glossary of AI techniques and their applications, please see page 26.

3.Contents Introduction: 2 Big prize, big impact 4 AI Impact Index 10 Realising the potential 20 Conclusion 22

4.Introduction: Getting down to what really counts Business leaders are asking: What impact These are the strategic questions we’ll be will AI have on my organisation, and is our addressing in a series of reports designed to help There’s a lot of business model threatened by AI disruption? enterprises create a clear and compelling business expectation And as these leaders look to capitalise on AI case for AI investment and development. While surrounding opportunities, they’re asking: Where should we there’s been a lot of research on the impact of artificial target investment, and what kind of capabilities automation, it’s only part of the story. In this new would enable us to perform better? Cutting series of PwC reports, we want to highlight how intelligence across all these considerations is how to build AI AI can enhance and augment what enterprises (AI). There’s in the responsible and transparent way needed can do, the value potential of which is as large, if also a to maintain the confidence of customers and not larger, than automation. significant wider stakeholders. amount of wariness. 2 Sizing the prize

5.The analysis carried out for this report gauges $15.7 trillion the economic potential for AI between now and 2030, including for regional economies and eight commercial sectors worldwide. Through our AI Impact Index, we also look at how improvements to personalisation/customisation, Game changer quality and functionality could boost value, choice and demand across nearly 300 use cases What comes through strongly from all the of AI, along with how quickly transformation analysis we’ve carried out for this report is just and disruption are likely to take hold. Other key how big a game changer AI is likely to be, and elements of the research include in-depth sector- how much value potential is up for grabs. AI could by-sector analyses. contribute up to $15.7 trillion1 to the global economy in 2030, more than the current output of In this opening report, we outline the regional China and India combined. Of this, $6.6 trillion is economies that are set to gain the most and likely to come from increased productivity and the three business areas with the greatest $9.1 trillion is likely to come from consumption- AI potential in each of eight sectors. Future side effects. reports will focus on specific sectors, along with While some markets, sectors and individual functional areas such as marketing, finance and businesses are more advanced than others, AI is talent management. We’ll also be setting out the still at a very early stage of development overall. detailed economic projections and, in partnership From a macroeconomic point of view, there are with Forbes magazine, publishing interviews with therefore opportunities for emerging markets to some of the business leaders at the forefront of AI. leapfrog more developed counterparts. And within your business sector, one of today’s start-ups or a business that hasn’t even been founded yet could be the market leader in ten years’ time. 1 $ denotes US dollars throughout, estimated values are expressed in real terms at 2016 prices (i.e. excluding the impact of general price inflation when looking ahead to 2030). What’s the real value of AI for your business and how can you capitalise? 3

6. AI touches almost every aspect of our lives. And it’s only just getting started. Big prize, big impact: Why AI matters How much is at stake and why should you take action? From the personal assistants in our mobile phones, to the profiling, customisation, and cyber protection that lie behind more and more of our commercial interactions, AI touches almost every aspect of our lives. And it’s only just getting started. According to our analysis, global GDP will be up to 14% higher in 2030 as a result of the accelerating development and take-up of AI – the equivalent of an additional $15.7 trillion. The economic impact of AI will be driven by: 1. Productivity gains from businesses automating processes (including use of robots and autonomous vehicles). 2. Productivity gains from businesses augmenting their existing labour force with AI technologies (assisted and augmented intelligence). 3. Increased consumer demand resulting from the availability of personalised and/or higher-quality AI-enhanced products and services. 4 Sizing the prize

7.How we gauged the impact and Over the past decade, almost all aspects of potential of AI how we work and how we live – from retail to manufacturing to healthcare – have become To estimate the impact and potential of AI, our team increasingly digitised. The internet and mobile conducted an ambitious, dual-phased top-down and technologies drove the first wave of digital, bottom-up analysis. In addition to drawing on input from known as the Internet of People. However, our extensive network of clients, and sector and functional analysis carried out by PwC’s AI specialists advisors within PwC, we’ve been working with our partners anticipates that the data generated from the Fraunhofer, a global leader in emerging technology research Internet of Things (IoT) will outstrip the data and development and Forbes. Together, we set out to identify generated by the Internet of People many times the most compelling examples of potential AI applications over. This increased data is already resulting across each sector’s value chain, and designed a framework in standardisation, which naturally leads to to assess the degree and pace of impact of each. In total, automation, and the personalisation of products we identified and rated nearly 300 use cases, which are and services, which is setting off the next wave of captured in our AI Impact Index. digital. AI will exploit the digital data from people and things to automate and assist in what we do Our Econometrics unit then used this bottom-up input as today, as well as find new ways of doing things part of their top-down analysis assessing AI’s impact on, that we’ve not imagined before. and the interactions between, key elements of the economy including labour, productivity, business and government. Productivity gains The models were informed by global economic datasets, In the near-term, the biggest potential economic extensive academic literature, and existing PwC work on uplift from AI is likely to come from improved automation. The analysis looked at the total economic productivity (see Figure 1). This includes impact of AI, accounting for increased productivity (which automation of routine tasks, augmenting may involve the displacement of some existing jobs), the employees’ capabilities and freeing them up to creation of new jobs, new products, and other effects. We’ll focus on more stimulating and higher value- be publishing an extended technical read out of these results adding work. Capital-intensive sectors such as later in the year. manufacturing and transport are likely to see the largest productivity gains from AI, given that For a more detailed methodology see page 27. many of their operational processes are highly susceptible to automation. Figure 1: Where 18000 will the value gains come from with AI? 16 Global GDP impact by effect of AI (£trillion) 14 12 $ trillion 10 8 6 4 2 0 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 Labour Productivity Personalisation Time Saved Quality As new technologies Labour productivity are gradually adopted and improvements are consumers respond to expected to account 58% of all improved products with GDP gains in 2030 for over 55% of all increased demand, the GDP gains from will come from share of impact from consumption side AI over the period product innovation 2017 – 2030. impacts. increases over time. Source: PwC analysis The impact on productivity could be competitively transformative – businesses that fail to adapt and adopt could quickly find themselves undercut on turnaround times as well as costs. They stand to lose a significant amount of their market share as a result. However, the potential of this initial phase of AI application mainly centres on enhancing what’s already being done, rather than creating too much that’s new. What’s the real value of AI for your business and how can you capitalise? 5

8.Increased consumer demand Healthcare, automotive and financial services are Eventually, the GDP uplift from product the sectors with the greatest potential for product enhancements and subsequent shifts in consumer enhancement and disruption due to AI according demand, behaviour and consumption emanating to our analysis. However, there is also significant from AI will overtake the productivity gains, potential for competitive advantage in particular potentially delivering more than $9 trillion areas of other sectors, ranging from on-demand of additional GDP in 2030. Consumers will be manufacturing to sharper content targeting mostly attracted to higher quality and more within entertainment we set out the business personalised products and services, but will areas with most AI potential in each sector in the also have the chance to make better use of their next section. time – think of what you could do if you no longer had to drive yourself to work, for example. In Some job displacement – but also new turn, increased consumption creates a virtuous employment opportunities cycle of more data touchpoints and hence more The adoption of ‘no-human-in-the-loop’ data, better insights, better products and hence technologies will mean that some posts will more consumption. inevitably become redundant, but others will be created by the shifts in productivity and consumer The consumer revolution set off by AI opens the demand emanating from AI, and through the way for massive disruption as both established value chain of AI itself. In addition to new types businesses and new entrants drive innovation of workers who will focus on thinking creatively and develop new business models. A key part about how AI can be developed and applied, a of the impact of AI will come from its ability to new set of personnel will be required to build, make the most of parallel developments such as maintain, operate, and regulate these emerging IoT connectivity2. technologies. For example, we will need the equivalent of air traffic controllers to control AI front-runners will have the advantage of the autonomous vehicles on the road. Same day superior customer insight. The immediate delivery and robotic packaging and warehousing competitive benefits include an improved ability are also resulting in more jobs for robots and for to tap into consumer preferences, tailor their humans. All of this will facilitate the creation of output to match these individual demands and, new jobs that would not have existed in a world in doing so, capture an ever bigger slice of the without AI. market. And the front-runners’ ability to shape product developments around this rich supply of Impact on different regions customer data will make it harder and harder for As Figure 2 highlights, some economies have slower moving competitors to keep pace and could the potential to gain more than others in both eventually make their advantage unassailable. absolute and relative terms. China and North We can already see this data-driven innovation America are likely to see the biggest impact, and differentiation in the way books, music, video though all economies should benefit. and entertainment are produced, distributed and consumed, resulting in new business models, new market leaders and the elimination of traditional players that fail to adapt quickly enough. 2 AI is the key to realising the promise of IoT as AI becomes an indispensable element of IoT solutions and the convergence of AI and IoT spur the development of ‘smart’ machines. We explore this further in ‘Leveraging the upcoming disruptions from AI and IoT’ (https:// 6 Sizing the prize

9.Figure 2: Which regions will gain the most from AI? Figure 1: Which regions gain the most from AI? Northern North Europe America China Total impact: Total impact: 9.9% of GDP 14.5% of GDP Southern ($1.8trillion) Europe Total impact: ($3.7trillion) 26.1% of GDP Total impact: ($7.0trillion) 11.5% of GDP ($0.7trillion) Developed Asia Latin Total impact: 10.4% of GDP America ($0.9trillion) Total impact: 5.4% of GDP ($0.5trillion) Africa, Oceania and other Asian markets Total impact: 5.6% of GDP ($1.2trillion) All regions of the Developing global economy will North America Total Europe and $10.7 Developed Asia countries will experience benefits and China stand to trillion 70% will also experience experience more from artificial see the biggest of the global significant economic modest increases intelligence. economic gains economic gains from AI due the much lower with AI enhancing impact enhancing rates of adoption of GDP by 26.1% GDP by 9.9% AI technologies expected. 14.5% 2030 11.5% 10.4% 2030 All GDP figures are reported in market exchange rate terms All GDP figures are reported in real 2016 prices, GDP baseline based on Market Exchange Rate Basis Source: PwC analysis Net effect of AI, not growth prediction 2. Our economic model results are compared to a Our results are generated using a large scale baseline of long-term steady state economic dynamic economic model of the global economy. growth. The baseline is constructed from The model is built on the Global Trade Analysis three key elements: population growth, Project (GTAP) database. GTAP provides detail on growth in the capital stock and technological the size of different economic sectors (57 in total) change. The assumed baseline rate of and how they trade with each other through their technological change is based on average supply chains. It gives this detail on a consistent historical trends. It’s very difficult to separate basis for 140 different countries. out how far AI will just help economies to achieve long-term average growth rates When considering the results, there are two (implying the contribution from existing important factors that you should take into account: technologies phase out over time) or simply be 1. Our results show the economic impact of AI additional to historical average growth rates only – our results may not show up directly (given that these will have factored in major into future economic growth figures, as there technological advances of earlier periods). will be many positive or negative forces that either amplify or cancel out the potential These two factors mean that our results should be effects of AI (e.g. shifts in global trade policy, interpreted as the potential ‘size of the economic financial booms and busts, major commodity prize’ associated with AI, as opposed to direct price changes, geopolitical shocks etc). estimates of future economic growth. What’s the real value of AI for your business and how can you capitalise? 7

10.North America China North America In North America, the potential uplift to GDP from The high proportion of Chinese GDP that comes AI will be amplified by the huge opportunities to from manufacturing heightens the potential is likely to see introduce more productive technologies, many uplift from introducing more productive the fastest of which are ready to be applied. And the gains technologies. It is likely to take some time to boost in the will be accelerated by the advanced technological build up the technology and expertise needed next few years. and consumer readiness for AI, along with the to implement these capabilities and therefore impact of rapid accumulation of assets – not just the GDP boost won’t be as rapid as the US. But in technology, but data touchpoints and the flows around ten years’ time, the productivity gains in of information and customer insight that come China could begin to pull ahead. with them. A key part of the value potential comes from the North America is likely to see the fastest boost in higher rate of capital re-investment within the the next few years. While the impact will still be Chinese economy compared to Europe and North strong from the middle of the 2020s, it probably America, as profits from Chinese businesses are won’t be quite as high as in the earlier years. fed into increasing AI capabilities and returns. One of the main reasons is that as productivity AI will also play an important part in the shift to in China begins to catch up with North America, a more consumer-oriented economy on the one this will stimulate exports of AI-enabled products hand and the move up the value chain into more from China to North America. sophisticated and high tech-driven manufacturing and commerce on the other. The focus and investment are amply demonstrated by the surge in AI patents filed in China3. An acceleration in talent development in areas such as analytics will be crucial in realising the potential gains from AI within the Chinese economy. Critical assumptions Our estimates reflect certain assumptions, which we will stress-test in our forthcoming detailed economic assessment. What happens if the pace of AI adoption is faster/slower in particular countries, for example? How does that affect the distribution of global growth? What happens if estimated changes in product quality do not materialise? A slowdown in the pace of AI uptake would delay the benefits that feed through to labour productivity. We see this as a key driver to both the timing and the overall impact of AI on GDP. We’re currently exploring the quantitative effect of several key scenarios. This includes examining alternate combinations of input parameters, as well as the timing of AI uptake. These sensitivity tests are designed to help better understand the risks around our results, while providing more insight into the parameters that drive the relationship between AI and economic growth. We plan to present the results of several scenarios in our detailed economic assessment. 3 China is now second behind the US in AI patent filings, a key indicator of long-term trends in technology. Source: ‘The Global Race for Artificial Intelligence – Comparison of Patenting Trends’, Wilson Center, 1 March 2017 ( global-race-for-artificial-intelligence-comparison-patenting-trends) 8 Sizing the prize

11.Figure 3: How quickly will AI impact productivity? North America and China are expected to witness the greatest GDP gains from AI increasing productivity, but the North America is expected to China will likely trajectory of the impact for the uptake AI technology two countries differs. realise the majority of AI benefits faster. more slowly but could see a large impact on GDP by 2030. How to respond? If your business is operating in one of the sectors Doing nothing is not a feasible option. It’s easy to or economies that is gearing up for fast adoption dismiss a lot of what’s said about AI as hype. Yet as our of AI, you’ll have to move quickly if you want analysis underlines, without decisive response, many to capitalise on the openings, and ensure your well established enterprises and even whole business business doesn’t lose out to faster-moving and more models are at risk of being rendered obsolete. cost-efficient competitors. In the short-term, many of the opportunities If you’re in one of the sectors or economies where and threats are likely to focus on productivity, the disruptive potential is lower and adoption efficiency and cost – the transformative phase. likely to be slower, there is still a significant If you’re the CEO of a transport and logistics challenge ahead – no sector or business is in any company, for example, you’re already seeing the way immune from the impact of AI. In fact, the impact of robots within packing and fulfilment potential for innovation and differentiation could operations. The bigger disruption will emerge be all the greater because fewer market players are when the sector switches to autonomous trucking. currently focusing on AI. The big question is how to Are you in a position to move ahead of your secure the talent, technology and access to data to competitors? What are the openings for vehicle make the most of this opportunity. manufacturers, technology companies and other potential new entrants to make inroads in your market? Could your business be at risk of becoming obsolete if you don’t move quickly enough? Automation in action An online insurer has leveraged an AI bot to automate the claims process from beginning to end. Instead of the days or even months it traditionally took to settle a claim, the bot is able to complete the entire pipeline from claims receipt, policy reference, fraud detection, payout and notification to customers in just three seconds. When rolled out at scale, this solution is poised to have a huge impact on the insurance industry. Source: PwC AI specialists What’s the real value of AI for your business and how can you capitalise? 9

12.AI Impact Index: Targeting and timing your investment AI is set to be the key source of transformation, The unique analysis within PwC’s AI Impact disruption and competitive advantage in today’s Index includes a rating for the potential to free fast changing economy. Drawing on the findings up time and enhance quality and personalisation. of our AI Impact Index, we look at how quickly We’ve used this analysis to create nearly 300 use change is coming and where your business can cases setting out the openings for innovation, the expect the greatest return. drivers, timings and current feasibility of market adoption, what could hold this up and how these In the research carried out for this report, we’ve barriers could be overcome. drilled down to the sector-by-sector and product- by-product impact of AI to enable your business to The areas with the biggest potential and target the opportunities, pinpoint the threats and associated timelines we outline at a high level judge how to address them. here are designed to help your business target investment in the short to medium term. Some aspects of change, such as robotic doctors, could be even more revolutionary, but are further off. 10 Sizing the prize

13.Figure 4: What’s the potential impact for your sector? Potential AI Consumption Impact % Adoption maturity – Near term (0-3 yr) Sector Subsector % Adoption maturity – Mid term (3-7 yr) % Adoption maturity – Long term (7+ yr) Healthcare 3.7 Providers/Health Services Healthcare: 3.7 Pharma/Life Sciences 37% 23% 40% Insurance Consumer Health Automotive 3.7 Aftermarket & Repair Automotive: 3.7 Component suppliers Personal Mobility as a Service 35% 47% 18% OEM Financing Financial Services 3.3 Financial Services: 3.3 Asset Wealth Management Banking and Capital 41% 59% 0% Insurance Transportation and Logistics 3.2 Transportation and Logistics: 3.2 Transportation 41% 41% 17% Logistics Technology, Communications and Entertainment 3.1 Technology, Communications and Entertainment: 3.1 Technology 47% 36% 17% Entertainment, Media and Communication Retail 3.0 Retail: 3.0 Consumer Products 54% 38% 8% Retail Energy 2.2 Energy: 2.2 Oil & Gas 39% 44% 17% Power & Utilities Manufacturing 2.2 Manufacturing: 2.2 Industrial manufacturing 14% 83% 3% Industrial Products/Raw Materials Grand Total 3.1 Scores based on PwC’s AI impact index evaluation. Potential scores range from 1-5, with 5 indicating the highest potential impact due to AI, and 1 being the lowest. What’s the real value of AI for your business and how can you capitalise? 11

14.Supporting diagnosis in one of the areas in healthcare with the biggest potential Healthcare Three areas with the biggest AI potential Longer-term potential: Robot doctors carrying • Supporting diagnosis in areas such as out diagnosis and treatment. detecting small variations from the baseline in patients’ health data or comparison with Barriers to overcome similar patients. It would be necessary to address concerns over • Early identification of potential pandemics and the privacy and protection of sensitive health tracking incidence of the disease to help data. The complexity of human biology and the prevent and contain its spread. need for further technological development also mean than some of the more advanced • Imaging diagnostics (radiology, pathology). applications may take time to reach their Consumer benefits potential and gain acceptance from patients, Faster and more accurate diagnoses and more healthcare providers and regulators. personalised treatment in the short and medium- term, which would pave the way for longer- High potential use case: Data-based term breakthroughs in areas such as intelligent diagnostic support implants. Ultimate benefits are improved health AI-powered diagnostics use the patient’s unique and lives saved. history as a baseline against which small deviations flag a possible health condition in Time saved need of further investigation and treatment. AI More effective prevention helps reduce the risk is initially likely to be adopted as an aid, rather of illness and hospitalisation. In turn, faster than replacement, for human physicians. It will detection and diagnosis would allow for earlier augment physicians’ diagnoses, but in the process intervention. also provide valuable insights for the AI to learn continuously and improve. This continuous Timing interaction between human physicians and the Ready to go: Medical insurance and smarter AI-powered diagnostics will enhance the accuracy scheduling (e.g. appointments and operations). of the systems and, over time, provide enough confidence for humans to delegate the task Medium-term potential: Data-driven entirely to the AI system to operate autonomously. diagnostics and virtual drug development. 12 Sizing the prize

15.Automotive Three areas with the biggest AI potential Medium-term potential: On-demand parts • Autonomous fleets for ride sharing. manufacturing and maintenance. • Semi-autonomous features such as driver assist. Longer-term potential: Engine monitoring and • Engine monitoring and predictive, predictive, autonomous maintenance. autonomous maintenance. Consumer benefit Barriers to overcome A machine to drive you around and ‘on-demand’ Technology still needs development – having an flexibility – for example a small model to get you autonomous vehicle perform safely under extreme through a city or a bigger and more powerful weather conditions might prove more challenging. vehicle to go away in for the weekend. Even if the technology is in place, it would need to gain consumer trust and regulatory acceptance. Time saved The average American spends nearly 300 hours High potential use case: Autonomous a year driving4 – think what you could with fleets for ride sharing that time if you didn’t have to spend it behind Autonomous fleets would enable travellers to the wheel. access the vehicle they need at that point, rather than having to make do with what they have or Timing pay for insurance and maintenance on a car that Ready to go: Automated driver assistance sits in the drive for much of the time. Most of systems (e.g. parking assist, lane centring, the necessary data is available and technology is adaptive cruise control etc.). advancing. However, businesses still need to win consumer trust. Predictive engine monitoring and maintenance technology is advancing. 4 American Automobile Association media release 8 September 2016 ( utes-driving-year/) What’s the real value of AI for your business and how can you capitalise? 13

16.Businesses can develop customised solutions rather than expecting consumers to sift through multiple options to find the one that’s appropriate. Financial services Three areas with the biggest AI potential developing customised solutions rather than • Personalised financial planning. expecting consumers to sift through multiple • Fraud detection and anti-money laundering. options to find the one that’s appropriate. • Process automation – not just back office Barriers to overcome functions, but customer facing operations Consumer trust and regulatory acceptance. as well. Consumer benefit High potential use case: Personalised More customised and holistic (e.g. health, wealth financial planning and retirement) solutions, which make money While human financial advice is costly and time- work harder (e.g. channelling surplus funds into consuming, AI developments such as robo-advice investment plans) and adapt as consumer needs have made it possible to develop customised change (e.g. change in income or new baby). investment solutions for mass market consumers in ways that would, until recently, only have been Timing available to high net worth clients. Finances are Ready to go: Robo-advice, automated insurance managed dynamically to match goals (e.g. saving underwriting and robotic process automation in for a mortgage) and optimise client’s available areas such as finance and compliance. funds, as asset managers become augmented and, in some cases, replaced by AI. The technology and Medium-term potential: Optimised product data is in place, though customer acceptance would design based on consumer sentiment and still need to increase to realise the full potential. preferences. Longer-term potential: Moving from anticipating what will happen and when in areas Assisted intelligence in action such as an insurable loss (predictive analytics) A financial services organisation used machine learning to proactively shaping the outcome (prescriptive to develop time-series clusters of their policyholder analytics) in areas such as reduced accident rates transactions. The machine learning solution helped or improved consumer outcomes. the company to identify common customer transaction patterns and better understand the key triggers driving Time saved variances. Combining policyholder data with external The information customers need to fully data on customer preferences, financial literacy, and understand financial position and plan for the other behavioural dimensions allowed the firm to better future is at their fingertips and adapts to changing predict which patterns would occur for each customer circumstances. Businesses can support this by persona. The organisation designed interventions around these insights, which opened the way for improved outcomes for both the customer and the company. Source: PwC AI specialists 14 Sizing the prize

17.Retail Three areas with the biggest AI potential Time saved • Personalised design and production. Less time exploring shelves, catalogues and • Anticipating customer demand – for example, websites to find the product that you want. retailers are beginning to use deep learning to predict customers’ orders in advance. Barriers to overcome Adapting design and production to this more agile • Inventory and delivery management. and tailored approach. Businesses also need to Consumer benefit strengthen trust over data usage and protection. On-demand customisation as the norm and greater availability of what you want, when and High potential use case: Personalised how you want it. design and production Instead of being produced uniformly, apparels Timing and consumables can be tailored on demand. If Ready to go: Product recommendation base on we look at fashion and clothing as an example, preferences. we could eventually move to fully interactive and customised design and supply in which AI created Medium-term potential: Fully customised mock-ups of garments are sold online, made in products. small batches using automated production, and subsequent changes are made to design based on Longer-term potential: Products that anticipate user feedback. demand from market signals. Retailers are beginning to use deep learning to predict customers’ orders in advance. What’s the real value of AI for your business and how can you capitalise? 15

18.Technology, communications and entertainment Three areas with the biggest AI potential Time saved • Media archiving and search – bringing Quicker and easier for consumers to choose what together diffuse content for recommendation. they want, reflecting their preferences and mood • Customised content creation (marketing, film, at the time. music, etc.). Barriers to overcome • Personalised marketing and advertising. Cutting through the noise when there is so much Consumer benefit data, much of it unstructured. Increasingly personalised content generation, recommendation and supply. High potential use case: Media archiving and search Timing We already have personalised content Ready to go: Content recommendation for recommendation within the entertainment sector. consumers. Yet there is now so much existing and newly generated (e.g. online video) content that it can Medium-term potential: Automated be difficult to tag, recommend and monetise. AI telemarketing capable of holding a real offers more efficient options for classification and conversation with the customer. archiving of this huge vault of assets, paving the way for more precise targeting and increased Longer-term potential: Use-case specific and revenue generation. individualised AI-created content. 16 Sizing the prize

19.Manufacturing Three areas with the biggest AI potential Time saved • Enhanced monitoring and auto-correction Faster response and fewer delays. of manufacturing processes. • Supply chain and production optimisation. Barriers to overcome Making the most of supply chain and production • On-demand production. opportunities requires all parties to have the Consumer benefit necessary technology and be ready to collaborate. Indirect benefits from more flexible, responsive Only the biggest and best-resourced suppliers and and custom-made manufacturing of goods, manufacturers are up to speed at present. with fewer delays, fewer defects and faster delivery. High potential use case: Enhanced monitoring and auto-correction Timing Self-learning monitoring makes the Ready to go: Greater automation of a large manufacturing process more predictable and number of production processes. controllable, reducing costly delays, defects or deviation from product specifications. There Medium-term potential: Intelligent is huge amount of data available right through automation in areas ranging from supply chain the manufacturing process, which allows for optimisation to more predictive scheduling. intelligent monitoring. Longer-term potential: Using prescriptive analytics in product design – solving problems and shaping outcomes, rather than simply predicting and responding to demand in product design. AI will facilitate more seamless integration of supply chain data, enabling anticipatory production and more efficient delivery of products to customers. What’s the real value of AI for your business and how can you capitalise? 17

20.Energy Three areas with the biggest AI potential Time saved • Smart metering – real-time information on More secure supply and fewer outages. energy usage, helping to reduce bills. • More efficient grid operation and storage. Barriers to overcome Technological development and high investment • Predictive infrastructure maintenance. requirements in some of the more advanced areas. Consumer benefit More efficient and cost-effective supply and usage High potential use case: Smart meters of energy. Smart meters help customers tailor their energy consumption and reduce costs. Greater usage Timing would also open up a massive source of data, Ready to go: Smart metering. which could pave the way for more customised tariffs and more efficient supply. Medium-term potential: Optimised power management. Longer-term potential: More efficient and consistent renewable energy supply in areas such as improved prediction and optimisation of wind power. 18 Sizing the prize

21.Transport and logistics Three areas with the biggest AI potential High potential use case: Traffic control • Autonomous trucking and delivery. and reduced congestion • Traffic control and reduced congestion. Autonomous trucking reduces costs by allowing for increased asset utilisation as 24/7 runtimes • Enhanced security. are possible. Moreover, the whole business Consumer benefit model of transport & logistics (T&L) might be Greater flexibility, customisation and choice in disrupted by new market entrants such as truck how goods and people move around and the manufacturers offering T&L and large online ability to get from A to B faster and more reliably. retailers vertically integrating their T&L. Timing Ready to go: Automated picking in warehouses. Medium-term potential: Traffic control. Augmented intelligence in action Longer-term potential: Autonomous trucking An automotive company developed a dynamic agent-based and delivery. model to simulate thousands of strategic scenarios for entering the ridesharing market. The model allowed key decision makers Time saved to test a variety of policy configurations in a virtual, risk-free Smart scheduling, fewer traffic jams and real-time simulated environment to help them understand the ultimate route adjustment to speed up transport. impact on market share and revenue over time, before actually making any decisions. Flight simulators allow pilots to test the Barriers to overcome impact of their decisions in a virtual environment to better Technology for autonomous fleets is still in prepare them for making decisions in flight, so why shouldn’t development. business executives do the same? Source: PwC AI specialists What’s the real value of AI for your business and how can you capitalise? 19

22.Realising the potential: What do you want from AI? Where do you begin? To prioritise your response, it’s important to How do you keep pace with change? map the key process flows to be automated and To prioritise decision flows to be augmented. What functions your response, 1/Work out what AI means for your contain high potential processes that could drive it’s important business near-term savings, for example? As data becomes The starting point for strategic evaluation is a scan the primary asset and source of intellectual to map the key of the technological developments and competitive property, what investments and changes would process flows pressures coming up within your sector, how enable you to capture more data and use it to be quickly they will arrive, and how you will respond. more productively? With this map in place, you automated and You can then identify the operational pain points can then develop the cost-benefit analysis for decision flows that automation and other AI techniques could automation and augmentation. address, what disruptive opportunities are opened to be up by the AI that’s available now, and what’s AI is applicable across all elements of the augmented. coming up on the horizon. value chain, which can lead to multiple silos of initiatives or confusion in finding a good starting 2/Prioritise your response point. Developing the insight, governance and In determining your strategic response, key organisational collaboration to pick your spot and questions include how can different AI options drive initiatives forward are therefore critical. help you to deliver your business goals and what is your appetite and readiness for change. Do you want to be an early adopter, fast follower or follower? Is your strategic objective for AI to transform your business or to disrupt your sector? AI provides the potential to enhance quality, personalisation, consistency and time saved, but it’s also important to consider the technological feasiblity of AI and the availability of the data needed to support AI. How are you planning to overcome barriers and accelerate innovation? 20 Sizing the prize

23.3/Make sure you have the right talent AI should therefore be managed with the same and culture, as well as technology discipline as any other technology enabled It’s important While investment in AI may seem expensive now, transformation. Key questions to ask while to prepare for PwC subject matter specialists anticipate that the building AI include: a hybrid costs will decline over the next ten years as the workforce in software becomes more commoditised. Eventually, • Have you considered the societal and ethical implications? which AI and we’ll move towards a free (or ‘freemium’ model) for simple activities, and a premium model for • How can you build stakeholder trust in the human beings business-differentiating services. While the solution? work side-by- enabling technology is likely to be increasingly • How can you build AI that can explain its logic side. commoditised, the supply of data and how it’s used so that a lay person can understand? are set to become the primary asset. • How can you build AI that is unbiased and transparent? To make the most effective use of this technology, it’s important to instil a data-driven culture that It’s important to put in place mechanisms to blends intuition and analytical insights with a focus source, cleanse and control key data inputs and on practical and actionable decisions across all levels. ensure data and AI management are integrated. Demand for data scientists, robotics engineers and Transparency is not only important in guarding other tech specialists is clearly growing. These against biases within the AI, but also helping to are in short supply, especially in less developed increase human understanding of what the AI can markets according to the interviews we carried do and how to use it most effectively. out with PwC’s data and analytics’ regional leaders, so it will be important to gear long-term We further explore business strategies for an training and development to these emerging AI world in ‘A strategist’s guide to artificial needs. As adoption of AI gathers pace, the value of intelligence’ (https://www.strategy-business. skills that can’t be replicated by machines is also com/article/A-Strategists-Guide-to-Artificial- increasing. These include creativity, leadership and Intelligence?gko=0abb5). emotional intelligence5. It’s important to prepare for a hybrid workforce in which AI and human beings work side-by-side. The challenge for your business isn’t just ensuring you have the right systems in place, but judging what role your people will play in this new model. People will need to be responsible for determining Autonomous intelligence in action the strategic application of AI and providing Entertainment industry consumers now have an unprecedented challenge and oversight to decisions. choice of movies, television, music and games. While this provides consumers with more opportunity to enjoy content 4/Build in appropriate governance specific to their unique tastes, they can experience ‘choice and control overload’ during their search. And worse, sometimes they make Trust and transparency are critical. In relation no choice at all! Video and music streaming companies have to autonomous vehicles, for example, AI requires begun using autonomous recommendation engines that combine people to trust their lives to a machine – that’s a segment trends, ratings and content similarity to personalise huge leap of faith for both passengers and public suggestions and engage customers. Engaging customers not only policymakers. Anything that goes wrong, be it a increases retention, but also allows companies to collect more malfunction or a crash, is headline news. And this data on individuals and improve the personalisation of offerings reputational risk applies to all forms of AI, not – creating a virtuous feedback loop that provides a significant just autonomous vehicles. Customer engagement competitive advantage. robots have been known to acquire biases through training or even manipulation, for example. Source: PwC AI specialists 5 We explore AI’s role in this creative process in ‘AI is already entertaining you’, strategy+business, 1 May 2017 ( article/AI-Is-Already-Entertaining-You?gko=dc252) What’s the real value of AI for your business and how can you capitalise? 21

24.Conclusion: The options for survival and success 22 Sizing the prize

25.Picture your market in five years’ time. How Get this right and creativity, collaboration and can you create the capabilities to compete? decision making within your organisation can The prize is all be empowered. You’ll have the potential to being far more As our analysis underlines, AI has the potential understand customer behaviour and anticipate capable, in a to fundamentally disrupt your market through and respond to their individual needs with a far more the creation of innovative new services and precision and foresight that have never been entirely new business models. We’ve already possible before. relevant way, seen the creative destruction of the first wave than your of digitisation. With the eruption of AI, some of The ultimate commercial potential of AI is doing business could the market leaders in ten, even five years’ time things that have never been done before, rather ever be without may be companies you’ve never heard of. In turn, than simply automating or accelerating existing the infinite some of today’s biggest commercial names could capabilities. Some of the strategic options that be struggling to sustain relevance or have even emerge won’t match past experience or gut possibilities disappeared altogether, if their response has been feelings. As a business leader, you may therefore of AI. too little or too late. have to take a leap of faith. The prize is being far more capable, in a far more relevant way, than Tomorrow’s market leaders are likely to be your business could ever be without the infinite exploring the possibilities and setting their possibilities of AI. strategies today. We believe there are four key questions your business should address if it wants to keep pace and capitalise on the opportunities: • How vulnerable is your business model to AI disruption? How soon will the change arrive? • Are there game-changing openings within your market and, if so, how can you take advantage? • Do you have the right talent, data and technology to help you understand and execute on the AI opportunities? • How can you build trust and transparency into your AI platforms and applications? What’s the real value of AI for your business and how can you capitalise? 23

26.Helping your business to make the most of AI PwC is already working with companies across Anand Rao each of the different sectors highlighted in this Global Leader of Artificial Intelligence, PwC report, to help them plan for and take advantage T: +1 (617) 530 4691 of AI to support their business strategy and E: improve performance. Twitter: @AnandSRao If you would be interested in a consultation about the potential within your business, please feel Gerard Verweij free to get in touch. Global & US Data & Analytics Leader, PwC US T: +1 (617) 530 7015 E: Twitter: @gverweij Euan Cameron Artificial Intelligence Leader, PwC UK T: +44 (0)20 7804 3554 E: Twitter: @euancameron55 24 Sizing the prize

27.Authors Dr. Anand S. Rao and Gerard Verweij We would also like to thank our academic and applied research partners at The authors would like to acknowledge the Fraunhofer, and especially Fraunhofer Big contribution of: Data Alliance, with contributions from the • Alan Morrison following institutes: • Barbara Lix • Fraunhofer Institute for Intelligent Analysis & Information Systems IAIS • Cathryn Marsh • Fraunhofer Institute for Applied Information • Chengyao Gu Technology FIT • Chris Curran • Fraunhofer Institute for Open Communication • Craig Scalise Systems FOKUS • Cristina Ampil • Fraunhofer Institute for Digital Media Technology IDMT • Edmond Lee • Fraunhofer Institute for Experimental Software • Hugh Dance Engineering IESE • John Ashworth • Fraunhofer Institute for Integrated Circuits IIS • John Hawksworth • Fraunhofer Institute for Material Flow and • John Sviokla Logistics IML • Jonathan Gillham • Fraunhofer Institute for Transportation and Infrastructure Systems IVI • Katherine Barnard Roberts • Fraunhofer Institute for Algorithms and • Lucy Rimmington Scientific Computing SCAI • Mark Paich • Fraunhofer Working Group for Supply Chain • Michael Schneider Services IIS-SCS • Pia Ramchandani • Fraunhofer Institute for Solar Energy Systems ISE • Shivanghi Jain 6 Examples include our article, ‘Will robots steal our jobs? The potential impact of automation on the UK and other major economies? ( What’s the real value of AI for your business and how can you capitalise? 25

28.Glossary AI consists of a number of areas, including but not limited to those below: Main AI areas Description Large-scale Machine Learning Design of learning algorithms, as well as scaling existing algorithms, to work with extremely large data sets. Deep Learning Model composed of inputs such as image or audio and several hidden layers of sub-models that serve as input for the next layer and ultimately an output or activation function. Natural Language Processing Algorithms that process human language input and convert it into understandable (NLP) representations. Collaborative Systems Models and algorithms to help develop autonomous systems that can work collaboratively with other systems and with humans. Computer Vision (Image The process of pulling relevant information from an image or sets of images for Analytics) advanced classification and analysis. Algorithmic Game Theory and Systems that address the economic and social computing dimensions of AI, such as Computational Social Choice how systems can handle potentially misaligned incentives, including self-interested human participants or firms and the automated AI-based agents representing them. Soft Robotics (Robotic Process Automation of repetitive tasks and common processes such as IT, customer servicing Automation) and sales without the need to transform existing IT system maps. 26 Sizing the prize

29.The basis for our analysis AI Impact Index To do this we used a combination of machine Our sector specialists worked with market learning and econometrics techniques. participants and our partners at Fraunhofer to • We followed a similar approach to the best identify and evaluate use cases across five criteria: practice we’ve used in job automation • Potential to enhance personalisation. research6. We used a machine learning • Potential to enhance quality (utility value). approach to assess the likelihood that a job can be automated based on the specific • Potential to enhance consistency. tasks involved. • Potential to save time for consumers. • We used econometric analysis to assess how • Availability of data to make these gains possible. much automation would impact productivity. Specific scoring parameters were derived for We use the KLEMS datasets (K-capital, each criterion. We also evaluated technological L-labour, E-energy, M-materials, and feasibility. The results helped us to gauge time S-purchased services) and quantified the to adoption, potential barriers and how they can relationship between the use of AI technologies be overcome. and productivity in different regions and industries. Economic analysis We used a multi-stage approach to first evaluate • We also drew insights from the AI Impact how much AI would proliferate in different Index and used academic literature on regions of the world and how much this would personalisation and marginal utility to affect jobs through automation, and second quantify the effect that AI will have on to assess how much this would impact the key consumers’ utility, choices available to them elements of the economy. and the amount of time they would save. • We brought all of this together into a global dynamic computable general equilibrium (CGE) model, an economic model of interactions between consumers, businesses and governments. Through this model, we analysed the total economic impact of AI focusing on the ‘net’ impact, accounting for the creation of new jobs, a boost to demand from product enhancements and other secondary effects. What’s the real value of AI for your business and how can you capitalise? 27