It is just over a year (November 2022) since ChatGPT propelled generative AI into the spotlight. Suddenly it is easy to create new text, images and code using natural language prompts. Unlike most AI before, generative AI is available right now to ordinary managers and workers in businesses and does not require a supporting IT or research department. So what does it mean for productivity growth?
According to Bain & Company (a consultancy), around a third of labour time in the economy could be automated using generative AI applications. The range varies from about 28% in retail to over 40% in professional services (see chart). In macro-economic terms, fully implemented, this would amount to a 50% increase in productivity, which is huge.
Of course, some of this may be hype, and it could happen over decades rather than a few years but it does seem to have the potential to boost productivity significantly. US labour productivity growth has run at around 1-1.5% pa this century, with Europe slower. An extra 1% per year would be a big deal.
The key to productivity growth is the speed of adoption and past experience suggests that the US will be the fastest to adopt this new technology as it has been with other technologies before. This could give the US economy a further productivity advantage in coming years.
Jobs changing but not mass unemployment
While AI brings fears of mass unemployment, the history of economic growth is of job roles disappearing or dwindling, from agricultural workers in the 19th century to typists and elevator operators in the 20th century. Generative AI will undoubtedly lead to some job roles going or fewer people required in that area. This will be painful for individuals, especially because many will have considered themselves relatively highly skilled and will have invested time and probably money to reach that position. But it would have to proceed extraordinarily rapidly to threaten mass unemployment.
AI also brings new capabilities such as customisation or ‘hyper-personalisation’ of sales efforts which will create the need for new employees to manage them and expand job opportunities. Understanding how to get the best out of generative AI has quickly become a job skill in itself, and also a dedicated job role.
The uses of AI
The technology seems most useful in marketing and sales, product and services development and customer and back-office support. Products like ChatGPT can write the first draft of marketing materials or, alternatively, edit a draft to produce a polished text in a particular style. The former can save an enormous amount of time while the latter can help people writing in a language that is not their first or those that missed out on learning to write well at school!
But there are risks too. The biggest risk may be the risk of inaccuracies. AI-generated text can present apparent certainties which in fact need checking. Another risk is litigation due to infringement of intellectual property since generative AI essentially borrows from the internet.
The generative AI space is changing rapidly with new products coming thick and fast. Meanwhile governments are looking to regulate in various ways, to protect copyright and to avoid abuses. Provided regulation does not shut it down (very unlikely), generative AI could make an appreciable difference to productivity. And, unlike many coming technologies it should start right away, if not already. That is very welcome.