Digital currencies: where to from here?
July 4, 2018
Here’s how to embed a culture of integrity in an organisation
July 10, 2018
Show all

Artificial intelligence is here to stay


Act fast or be left behind

The traditional levers of economic progress – capital investment and labour – are no longer able to sustain the growth in GDP enjoyed in previous decades in most developed economies. With the recent convergence of technology, economies are entering a new era where artificial intelligence (AI) has the potential to overcome the physical limitations of capital and labour, and open up new sources of value and growth. Of late, communities, businesses and even governments have become increasingly concerned about how they can capitalise on AI, and if their comparative advantage is threatened by AI disruption. They are looking to AI capabilities which can allow them to outperform their peers, improve productivity, and even enhance the quality of their goods and services.

If it ain’t broken, why fix it?

Many jobs that used to exist a century and a half ago no longer exist today. To illustrate, in 1870, almost 50% of the United States (US)’ workers were in the agricultural sector, supplying food to the country. Fast forward to today – agriculture employs shy of 2% of the nation’s workers, and surprisingly, American food production outstrips domestic demand. We can thus see how technological innovations, like the self-driving tractors we see today, have improved the productivity of the sector and subsequently, improved standards of living.

Generally, technology improves productivity by reducing the number of hours required to generate a unit of output. Labour productivity improvements normally result in increases in average wages, thus lowering the number of work hours for the worker. Empirically, over the last 65 years, a large part of developed economies has experienced a significant decline in the average annual hours worked per worker.

Figure 1 Average annual hours worked per worker

Source: OECD Stat

Enter the 2nd stage of the industrial revolution

Indeed, technology is one of the key forces of the growth in productivity. In particular, a study for the Centre of Economic Performance at the London School of Economics concluded that industrial robotic automation raised GDP and labour productivity by 0.37 and 0.36 percentage points annually between 1993 and 2007 for the 17 countries it studied.

However, there has been a stark fall in the potential of capital investment, combined with labour, to boost the rate of growth currently experienced. World GDP growth has been steadily declining since its peak of 5.4% since 2011.

Figure 2 Labour productivity growth in G7 economies

Source: Conference Board, Total Economy Database, CEA calculations

As such, both developing and developed economies may have to leverage on the capabilities of AI to overcome these incumbent limitations, and to explore new avenues of growth.

Leaders and policymakers should look to AI as a game changer, rather than as a temporary hype in the market. AI’s presence is stronger now than ever before. The Indxx Global Robotics & Artificial Intelligence Thematic Index, designed to track the performance of companies listed in developed markets that are expected to benefit from the increased adoption of robotics and AI, rose almost 90% between 2015 and 2017. This increased inflow of funds and interest among investors in AI and robotics has been growing for the better part of the decade.

Implications on GDP

Shifting our lens onto GDP prints instead, we observed that G4 GDP growth rates have been steadily falling for the past two decades. Thus, we feel that AI can offer the developed economies fuel to improve productivity and streamline processes to once again drive economic growth higher.

Figure 3 Growth in GDP per capita

Source: OECD Stat

To have a closer glimpse of AI’s estimated effect on global GDP as well as GDP in developed economies, we took a dive into various white papers from the likes of PwC, Accenture (“Why Artificial Intelligence is the Future of Growth 2017”) and Frontier Economics. We found their methodology to be interesting and present some of these effects below.

Figure 4 Impact of AI on global GDP (£ trillion)

Source: PwC Analysis

The first paper we delved into was a research report by PwC (“Sizing the Prize”, 2017). To estimate the potential and impact of AI, their team sourced for input from their database of customers, as well as functional and sector advisors within the firm. Many insights were drawn from the extensive research but we choose to focus mostly on the potential impact of AI on GDP, dividing the impact into consumption (demand-side) and productivity (supply-side) effects.

On the data front, we saw that enhancements in productivity of labour were slated to take up more than 55% of overall GDP improvement from AI from 2017 to 2030. As seen from the curve in Figure 4, as new technology is slowly taken up and implemented by mainstream consumers, there will be an uptick in demand, thus increasing the proportion of the impact from product advancements. Additionally, as the impact from increased productivity starts to taper off from 2025, demand-side effects would have the lion’s share of the impact of AI on GDP, hitting around an estimated of 58% by 2030.

Figure 5 Projected GDP of G7 economies in 2035

Source: Accenture, Frontier Economics

Another paper, a collaboration by Accenture and Frontier Economics, helped us understand the impacts of AI on 12 developed economies, of which we choose to present the largest six (Figure 5). These 12 developed markets, which contribute more than half of the world’s output, aided us in realising that AI has the capabilities to double annual GDP rates, providing a solution to the slowing growth rates of late.

To model the economic potential of AI on these various economies, a baseline and expected economic growth factoring the impacts of AI were included. Furthermore, a buffer till 2035 was used in the analysis to take into account the time for the impacts of AI to filter through. Clearly, AI’s modelled economic benefit on the US is substantial, with the potential to boost GDP growth to 4.6% by 2035. Also, Germany and Japan’s economies stand to benefit significantly, with the potential to double and triple their GDP growth rates by 2035, compared to a baseline scenario without AI enhancements. Elsewhere, the projected effects on GDP were also seen with more laggard economies such as Italy (80%), Spain (47%) and Belgium (69%), albeit with duller effect. This is due to the mismatch of technological advancement and public investment in these laggard economies.

These estimates also allow us to see how quickly these regions can double in size with the help of AI. Particularly for Japan, the baseline effect (no AI intervention) would require around 80 years for its economy to double in size. However, with AI, it is estimated that Japan’s GDP could double in a mere 25 years. The US is even quicker in this aspect, potentially boasting a doubling in size is just 10 to 15 years with the help of AI intervention.

This all points to one indisputable conclusion – AI is here to stay, and nations will have to act fast, or risk being left behind.

Selena Ling is Head of Treasury Research & Strategy, OCBC Bank.

This article was reproduced by the Institute of Singapore Chartered Accountants, having first been published in the OCBC Global Outlook 2018. You can also read the article here