As debate continues as to what AI means for global economics and investment, the world's largest wealth management house weighs in on the topic.
The rise of artificial intelligence may change how people regard inflation and, at one level, render the concept of a general price level “meaningless,” UBS argues in a note about AI and macroeconomics.
While debate exists on whether generative AI tools such as ChatGPT count as “intelligence” in a meaningful sense yet (see a related editorial here), the ascent of these offerings has grabbed attention around the world.
In its “TechGPT” edition, UBS along with chief economist Paul Donovan, said there are potential macroeconomic consequences from AI that unfold. That includes inflation – a hot topic in the US and other countries.
“One application of AI may be to more readily identify an individual’s determination to buy a product, which would allow the seller to charge the highest possible price that the buyer will be prepared to pay,” the bank said. “Modern calculations of inflation assume a universal price – indeed, they exclude prices generated by schemes that give certain groups a special price discount. AI may make the idea of a universal price an increasingly meaningless concept, in our view.”
“In theory, if AI improves efficiency, there should be a reduction in inflation. Efficiency gains may go into higher profit margins, but over time, economists would expect some of those profits to be competed away – and the competition involved suggests a lower rate of inflation,” UBS continued. “However, AI should also change relative prices. Thus, the value placed on certain workers or certain commodities may adjust as AI makes some things possible that were previously uneconomic. Within the idea of greater efficiency, there is also the concept of increased volatility for individual prices. Over time, we think AI may make the concept of economy-wide inflation meaningless.”
Statistically, the bank tried to work out how the inflation numbers pan out depending on specific business sectors.
“Within tech, software and semiconductors (both carrying operating margins of around 30 per cent) enjoy better pricing power, whereas hardware and IT services with only low to mid-teen margins are mostly at risk if AI significantly changes the way goods and services are priced,” UBS said.
Along with its peers, UBS its trying to work out not just the macroeconomic effect of AI but also how it affects the wealth management value chain. Private banks and wealth managers also have to confront how such technology changes the jobs of their staff. The CFA Institute, for example, which is a major accreditation and professional training organisation, has decided to bring AI into its field.
“The current success of many generative AI applications like ChatGPT, Bard and Midjourney is redefining the way corporates, consumers and governments approach technology, in our view,” UBS continued.
“While the full impact on GDP growth is not clear, we believe software companies with first-mover advantages in AI that can drive productivity should be clear winners,” it said.
“As more routine tasks are automated, worker productivity should improve. In this, AI is no different from almost every other technological innovation over the centuries. As with previous technological changes, the emphasis should be less about the technology itself and more about how businesses adapt to implement the technology. Maximising productivity may require a rethinking of business models, not just tweaking existing practices to accommodate AI,” it said.
However, UBS said the growth implications of AI are less clear, arguing that gross domestic product (GDP), as a measurement of improving living standards, is a rather “antiquated concept.”
“AI may lead to increased production (which is GDP-positive). It might also focus on more efficient production – and while efficiency increases living standards, it does not necessarily increase GDP. For instance, applying AI appropriately might result in unchanged output but increased leisure time. That is a positive for input economic measures of growth, but not output economic measures of growth like GDP,” it said.
Jobs lost and created
UBS said industries such as IT services which depend the most on human capital (employee costs are greater than 40 per cent of sales) are at a risk of disruption from significant automation.
The bank grappled with worries that AI will erase whole fields of work and make humans superfluous.
“Ever since the First Industrial Revolution, every technological change is often seen as reducing employment. Every time the argument is wrong, in our view. It is certainly true that some jobs will be lost to AI. It is important to look at the individual tasks someone performs in their job. The simplistic rule is `if half your job can be automated, change your job. If less than half your job can be automated, your job will change’. Clearly, there will be jobs in the first category that technology will render obsolete, as has happened in the past. Few offices today have `typing pools’, for example,” it said. “AI is also likely to create new jobs that have not previously been thought of. Roughly 10 per cent of the labour roles that exist at the end of any decade did not exist at the start of the decade. This is either because technology creates new opportunities that were not previously possible, or because it lowers barriers to entry in existing professions. A parallel example is the increase in employment in the entertainment industry, as social media and streaming collapsed the barriers to entry in music and film.”
UBS said such arguments show that labour market flexibility is crucial – retraining for jobs that are obsolete or accepting the need to adapt, removing legal and social barriers, etc.