Terminated or turbocharged? Analysing AI’s potential impact on jobs, industries and growth

Key takeaways

• Demand and future revenue growth from AI-related hardware, software and services could be strong
• Concerns over smart robots taking jobs look unfounded as yet, although inequalities could play out across different lines
• AI should boost industry productivity – improving personalisation and product experience, and modelling complex systems better to generate efficiency savings. Investors expect start-ups, big tech and advanced semiconductor firms to capture value
• AI applications also face constraints: computing power, regulatory, legal and moral issues, and the sometimes lengthy timescale to commercialise new technologies.

We consider whether AI can boost productivity, whether new advancements threaten jobs, which industries will capture its value, and which will be transformed by its applications.

Smart robots – sci-fi fear or impending reality? The advent of artificial intelligence (AI), a technology that enables machines to perform tasks typically requiring human brains, dates back to the 1950s. But the launch of AI content-creator ChatGPT in November 2022 has itself unleashed a flood of content on the transformative power of AI and machine learning. Google launched its own, rival AI chatbot called Bard on 21 March. So what is new? ChatGPT uses ‘generative’ language processing algorithms based on ‘foundation models’: in other words it learns from a vast amount of data to create new texts, and it improves over time. Other tools could do the same with images, audio and video. Such advanced ‘deep learning’ models have huge potential across many applications (including in robots), particularly now our society creates exponential amounts of data that traditional computing techniques struggle to digest and exploit. Ubiquitous smartphones, internet communication and small, fast computer chips, as well as the proliferation of cameras and sensors, is another tailwind for AI applications. Demand appears strong. From its launch four months ago by a start-up firm, ChatGPT now has 100mn+ active users. Market intelligence firm IDC forecasts global revenues for the AI market – including hardware, software and service sales – will rise 19% per year from now until 2026.

Rage against the machine

Will AI-enabled applications take jobs from humans – or increase inequality by taking them from unskilled workers? The current backlash against globalisation reminds us that shifting labour patterns can have big political consequences. Yet a 2022 Massachusetts Institute of Technology study of Finnish manufacturing firms’ concluded that their use of advanced technologies had actually led to increases in employment. Research by South Korea’s central bank also found that robotisation did not decrease overall job vacancies in the service sector and non-routine jobs. Of course, many robots are mechanical rather than AI-enabled. And AI’s application extends far beyond robotics. Yet one interesting feature of new AI applications like ChatGPT is that the focus is on skilled rather than repetitive tasks, including parts of jobs done today by programmers, academics, lawyers and writers. Society might ask where humans’ relative advantages will be if AI takes over cognitive abilities.

Still, even with the use of AI, machines will still require humans to programme them and to check the accuracy, reliability and impact of their work. For now, smart robots are expensive, and struggle to perform ‘human’ tasks that blend cognitive skills with movement – although of course, further improvements should make them more economical and easier to deploy. Interestingly, fears over rising inequality may be felt differently in a new age of AI, notably by countries less able to supply the computing infrastructure to support them.

Keeping the fAIth

Could new AI applications unleash a productivity boom, delivering a shot in the arm to western economies? ‘Intelligent’ robots could take over more dangerous jobs, or those that people are less willing to do: working in power plants, war-zones, laboratories or abattoirs, stacking shelves or fulfilling orders, freeing up humans to do more rewarding tasks without eliminating jobs overall.

The contribution of AI-enabled technologies may be coming at a good time: western societies are short of workers. Unemployment in the US and Europe is near record lows. In the US and UK many people appear to have left the workforce permanently. Ageing populations will reduce it further: it is no coincidence that some of the countries with the highest penetration of robots today are also those that are ageing fastest. Pressure on companies to shift supply chains closer to home lowers their ability to source cheap labour wherever it can be found. Meanwhile, new technologies are sorely needed to tackle existential problems such as the climate crisis, and how to feed a growing population.

Perhaps the advent of AI will create entirely new jobs. Demand for individuals who can develop complex AI algorithms is already rising. It can be hard to predict the impact of new technologies. Cars and smartphones unleashed vast global demand for hitherto unimaginable products. If AI results in more automated video and image generation, then consumption of games, VR simulations and other visual media could rise. Or AI may create markets for entirely new products and services.

Where is value being created?

Who will emerge as the AI winners? New technologies are built on increased computing power, which is good news for makers of the hardware that underpins them: connectivity, accelerators, and memory components used in data centres, e.g. advanced semiconductor chip firms. The second group of companies likely to reap the rewards are big technology firms and software specialists. New AI applications require big research and development budgets, reams of data, and multiple users to refine advancements with their feedback. Advances in AI can often be quickly translated into consumer-facing tech products like better search engines, chat bots, translation tools, virtual and voice assistants. Microsoft has invested USD 10 billion in ChatGPT maker OpenAI, and will use its models across its consumer and enterprise products. Microsoft’s biggest rivals are reportedly working on similar technologies. Meta Platforms, formerly Facebook, recently said its single largest investment would be in advancing AI and building it into all of its products. In early March, Google and the Technical University of Berlin unveiled an AI robot that can undertake complex tasks combining language, vision and reacting to its environment. Today’s AI race could also play out along geopolitical lines, pitting giant US and Chinese firms against each other in a bid to gain the upper hand in potentially critical new technologies. Meanwhile, investors hope that advances from small cap firms, too, will create value: flows into venture capital-backed AI start-ups hit USD 1.37 billion in 2022, estimates data provider Pitchbook, almost as much as in the previous five years combined.

Personalisation and productivity gains

Beyond this tech focus, advances in AI – if they play out as many hope – will likely have broad-based implications across sectors. Indeed, they are already having a big impact, focused in two areas: improving the experience and personalisation of products and services, and enabling firms to make efficiency savings, often through improved modelling of complex systems. Take the first example. With the help of well-developed recommendation systems, big media companies and asset managers have become more adept at targeting TV, music and investment ideas to their subscribers and clients. Retailers have a greater understanding of consumer preferences – from their fashion favourites right down to tracking which clothes might fit their individual measurements. The ability to search images, and better virtual assistants, are improving the online shopping experience.

The second example – productivity and efficiency savings – has many applications. Energy firms are using machine learning algorithms to automatically interpret seismic data in the search for new oilfields, and to predict when mechanical equipment will need maintenance. Banks are modelling credit dynamics better across customer populations, in the hope of improving default predictions. Insurers are using ‘cognitive computing’ to assess the impacts of simple accident claims and reduce settlement periods. AI can improve the management of supply chains, inventories and invoicing across industries.

Healthcare firms provide a particularly interesting insight into the many uses of AI. The sector was comparatively late to digitalise, but the sheer mass of healthcare data, the rising incidence of chronic diseases and the need for swift vaccine development during the Covid-19 pandemic have all accelerated the use of machine learning. Today, every stage of a drug’s lifecycle – from the choice of molecules to develop and the design of clinical trials, to commercialisation – can be improved. Epidemiologists use machine learning to model disease outbreaks. ‘Smart’ medical devices can help diabetics monitor their blood sugar, and clinicians make more informed diagnoses. US regulators approved the first autonomous AI diagnostic system in 2018, to diagnose diabetic retinopathy directly from patient scans.