|Practice area:||Data & technology|
|Published:||19 January, 2021|
|Keywords:||Artificial Intelligence LETech|
Business interest in AI has increased massively over the last decade, and with it the fear that businesses that don’t adopt AI will be left behind and perish. Yet, businesses often struggle with the implementation; they lack the skills and knowledge, and often the case for where AI could add value is not clear.
What is artificial intelligence?
Artificial intelligence (AI) is a branch of computer science that is concerned with building ‘smart’ or ‘intelligent’ machines. AI is a vast field encompassing a wide range of different subfields, including: machine learning, concerned with developing algorithms that learn from data, natural language processing, concerned with reading and understanding text; speech recognition, concerned with understanding human speech; computer vision, concerned with understanding images and videos; and many more.
AI is part of a wider movement towards digitalisation, the trend towards wider use of digital technologies, increased connectivity, smarter products and machines. Digitalisation encompasses developments in a range of fields including purely digital technologies such as AI; but often also digital technologies working together with physical technologies such as sensor technologies, telecommunication, the internet of things, robotics and autonomous systems, and so on.
The hype and history of AI
AI has seen a massive increase in interest more recently. Analysis of Google Trends data shows that interest in AI has grown by more than 70% over the last decade. The last few years have also seen an ever-increasing flurry of news articles proclaiming that businesses not investing in AI will be left behind and perish.
Interest in AI in the UK over the last decade (Index, 2010=100)
Note: The figure shows the average interest, as measured by the volume of Google searches in the UK, in artificial intelligence over time relative to 2010. Keywords used: “Artificial Intelligence” and “AI”.
Source: London Economics analysis of Google Trends data
While many may only have become aware of AI recently, it is not a new concept. First developments in AI can be traced back to the ‘50s, when the first artificial intelligence program was presented in 1956 at the Dartmouth Summer Research Project on Artificial Intelligence (Anyoha, 2017). AI has also been used in every-day applications such as spam filtering and optical character recognition (OCR) for a long time.
However, what has changed is that, over the last decade or so, we have seen a massive increase in the amount of computing capacity available, with mobile phones now being more powerful than even the computers used by NASA to place astronauts on the moon (Todorov, 2014). This increase in computing capacity has enabled AI to be used at a scale like never before, enabling a wide range of use cases which were previously simply computationally infeasible. This increase in computing capacity has coincided with increasing amounts of data availability (from ubiquitous sensors, digital transactions, online activity, etc.) that make the application of AI both a practical necessity and commercially attractive.
The benefits of AI: Why the hype?
AI, and digitalisation more widely, has the potential to bring significant benefits to the UK economy. While estimates of potential future benefits naturally carry significant uncertainty, several studies suggest that the impacts AI could bring are sizeable (see for example PwC (2017), ITU (2018), McKinsey (2018, 2019)). For the UK, the potential benefits AI could deliver to the economy have been estimated, in 2016, to be in the region of £600 billion ($814 billion) by 2035 (Accenture, 2016).
Benefits of AI to firms include cost reductions and efficiency improvements arising from substitution of work from humans and less capable digital or analogue systems to AI; as well as effects resulting from better decision-making and freed up labour time for other (higher value-add) tasks. In addition, AI can help firms transform their products and services, for example through increased personalisation, as well as improve their business services and consumer interaction. This in turn also creates further benefits for consumers from improved products and services, better customer service and more personalised and targeted products, services and interactions. An especially pertinent application of AI at the time of writing is its use in tracking the spread of COVID-19 (Brice 2020); it is of course also used in discovery research / R&D more widely.
Indeed, AI is already used in a wide range of tasks across various sectors. For example, AI chatbots have already become a first point of contacts for many businesses. In leading law firms, AI-enabled software such as Luminance and Kira helps lawyers with document research and review. In some advanced factories, AI is used to predict bottlenecks in production. In agriculture, AI can already bed used for crop and soil monitoring. In finance, AI can help with fraud detection and risk prediction. Many more examples where AI is already used exist.
If AI is so great, why isn’t everyone already using it?
Despite existing use-cases and evidence of the benefits AI can bring, many businesses are struggling to embed AI in their core business model. This is particularly the case among smaller firms, with a survey by the National Bureau of Economic Research (Zolas et. al., 2020) finding that existing estimates of AI adoption are massively overstated due to the over-representation of large firms in the sampling frames. The survey, which was focused on the US, found that the share of firms using any form of AI was less than 9% in 2016. In short, AI adoption is heavily skewed towards large firms, with the vast majority of small firms not using AI at all. While AI adoption continues to spread, wider diffusion of AI in the UK is still at relatively early stages (McKinsey, 2019).
A barrier that immediately comes to mind is the lack of data scientists and other staff with the right skills to integrate AI in core business functions. Indeed, a lack of skills has long been recognised as a key obstacle to AI adoption (see, e.g., McKinsey (2018), Deloitte (2020), etc). However, while lack of skills is undoubtedly a challenge, a perhaps more fundamental challenge hindering adoption, and indeed business model transformation, is a lack of understanding, by business leaders, of how AI can be utilised within the core business and a lack of institutional support.
In response to a recent survey we undertook for the AI for Services report, 81% of stakeholders in the high value services sector agreed that a limited understanding of AI and data technologies and the benefits it can bring was a key challenge to the further adoption of AI in the services sector. This is mirrored in research by Microsoft (Rosenshine, 2020) that found that two-thirds of business leaders do not understand how AI arrives at conclusions. Moreover, 71% agreed that a lack of ownership and commitment to AI by senior management was a key challenge (London Economics, 2020).
Given this lack of understanding and skills, it is not surprising that many companies lack the clear AI and data strategy necessary to fully exploit the benefits of AI. In practice, AI is also often constrained by a lack of data standards, and by the presence of legacy systems and data ‘silos’.
AI is not always the only or indeed best solution:
At the same time, the hype surrounding AI means that companies can be left wondering whether they need to adopt or ‘be left behind’ yet have no clear idea of where AI could improve their core business. Adopting for adoption’s sake is almost always a bad idea. AI is not a magic pill, and many of the cutting-edge developments may not be right for every business. Indeed, in many cases much simpler solutions may be more appropriate, require less time, and ultimately deliver more immediate value. Therefore, rather than worrying about missed opportunities, businesses should examine the needs of their business and work backwards to the most appropriate solution.
Investing in a good data infrastructure is another action businesses considering whether they should adopt AI should take prior to adoption. Good data is the key to good AI; AI requires a good data infrastructure to yield its full potential. Therefore, this is a good way to ensure the business is ready to adopt AI should the need arise, while being able to derive benefit from available data in the short term.
Adopt or perish: Is AI a magic bullet?
AI has the potential to bring significant benefits to UK companies and the wider UK economy. The hype around AI means that news articles warning businesses will fail unless they adopt AI now have sprung up everywhere. However, in reality, companies often struggle with AI adoption. For small and medium-sized enterprises in particular AI, for now, is not the magic bullet it perhaps appears to be. Use cases within the core business are often not clear, and the value added is not obvious. Rather than adopting for adoption’s sake, businesses should focus on the challenges they are facing and consider solutions that will help overcome those challenges. Simpler solutions may often be more relevant, quicker to implement, and bring more value than the latest trend in AI. Moreover, businesses considering adopting in the future should focus on and invest in their data infrastructure; this will place them in a good position should they take the plunge to adopt in the future.
Author: Daniel Herr
Anyoha R. (2017). Can Machines Think? Harvard University. Available at: http://sitn.hms.harvard.edu/flash/2017/history-artificial-intelligence/ [accessed 2020-01-18]
Brice, P. (2020). Putting AI to work against COVID-19. University of Cambridge. Available at: https://www.phgfoundation.org/blog/Putting-AI-to-work-against-COVID-19 [accessed 2020-01-18]
Deloitte Insights (2020). The AI talent shortage isn’t over yet: Even mature AI adopters are challenged by skill gaps. Available at: https://cloudblogs.microsoft.com/industry-blog/en-gb/cross-industry/2020/01/08/3-ways-organisations-can-use-ai-in-a-responsible-way/ [accessed 2020-01-18]
ITU (2018). Assessing the Economic Impact of Artificial Intelligence. ITUTrends: Emerging trends in ICTs. Issue Paper No. 1. September 2018. Available at: https://www.itu.int/dms_pub/itu-s/opb/gen/S-GEN-ISSUEPAPER-2018-1-PDF-E.pdf [accessed 2020-01-18]
McKinsey Global Institute (2018). Notes from the AI frontier: Modelling the impact of AI on the world economy. Discussion Paper. Available at: https://www.mckinsey.com/~/media/McKinsey/Featured%20Insights/Artificial%20Intelligence/Notes%20from%20the%20frontier%20Modeling%20the%20impact%20of%20AI%20on%20the%20world%20economy/MGI-Notes-from-the-AI-frontier-Modeling-the-impact-of-AI-on-the-world-economy-September-2018.ashx [accessed 2020-01-18]
McKinsey Global Institute. (2019). Notes from the AI frontier: tackling Europe’s gap in digital and AI. Discussion Paper. Available at: https://www.mckinsey.com/~/media/mckinsey/featured%20insights/artificial%20intelligence/tackling%20europes%20gap%20in%20digital%20and%20ai/mgi-tackling-europes-gap-in-digital-and-ai-feb-2019-vf.ashx [accessed 2020-01-18]
PricewaterhouseCoopers (2017). The economic impact of artificial intelligence on the UK economy. Available at: https://www.pwc.co.uk/economic-services/assets/ai-uk-report-v2.pdf [accessed 2020-01-18]
Rosenshine, K. (2020). 3 ways organisations can use AI in a responsible way. https://cloudblogs.microsoft.com/industry-blog/en-gb/cross-industry/2020/01/08/3-ways-organisations-can-use-ai-in-a-responsible-way/ Available at: https://cloudblogs.microsoft.com/industry-blog/en-gb/cross-industry/2020/01/08/3-ways-organisations-can-use-ai-in-a-responsible-way/ [accessed 2020-01-18]
Todorov, T. (2014). A modern smartphone or a vintage supercomputer: which is more powerful? phoneArena.com. Available at: https://www.phonearena.com/news/A-modern-smartphone-or-a-vintage-supercomputer-which-is-more-powerful_id57149 [accessed 2020-01-18]
Zolas, N. et. al. (2020) Advanced Technologies Adoption and Use by U.S. Firms: Evidence from the Annual Business Survey. National Bureau of Economic Research. Available at: https://conference.nber.org/conf_papers/f138039.pdf [accessed 2020-01-18]
 Converted to Sterling using the Bank of England average annual spot exchange rate for 2016.
 78% of stakeholders responding to our survey for the AI for Services report agreed that a lack of a clear AI and data strategy was a key challenge (London Economics, 2020).