The old world is dying

Gramsci used this line to describe periods of deep structural transition, where existing systems lose legitimacy before workable alternatives are fully formed. It is a useful way to think about the current moment in artificial intelligence.

This article is the first in a three-part series, exploring how AI development is in a race in which labour markets and institutions must keep up and adapt. As firms rapidly adopt new tools, increased productivity is promised. But how the transition is managed will dictate the distribution of the benefits from AI between capital and profits,  and labour and wages.

Regardless of whether one is optimistic or sceptical about AI’s long-term potential, its presence is already reshaping how work is done. Workers, businesses, academic institutions, and governments are confronting changes to established norms and practices. 

From universities grappling with students using AI to support essay writing, through to the Ministry of Education marking exams with AI, the technology has moved from the margins into institutional systems, while firms integrate it into their operations.

According to McKinsey’s latest global survey on the state of AI, around 88 percent of organisations now use AI in at least one business function, up from 55 percent just three years earlier. Nine out of 10 surveyed organisations report regular use of generative AI, particularly in areas such as marketing, software development, operations, and professional services. 

With the big four tech companies investing between $800 and $900 billion of capex in 2026, more than $1 trillion in 2027, there are great expectations within the industry, and the wider economy, for the economic potential that AI will provide. AI is expected to be the next big thing, allowing these tech giants to capitalise on early adoption and investment.

But beneath this investment boom, driven by potential returns, AI’s potential ultimately lies in automating tasks. While it may reduce reliance on hard-to-hire staff, or hard-to-find expertise, the risk is that at the margin AI leads to less investment into skills by substituting away from labour.

AI is changing how work is done. How the returns to AI are shared depends on our management of the transition, and how benefits will be shared between firms and employees. 

AI allows for the standardisation of tasks and reduces dependence on individual workers, affecting bargaining relationships for compensation. As the distribution of gains tilts toward capital, increased productivity would translate less into marginal wage growth (all other things equal).

Studies out of Stanford have shown that there is a growing reduction in early job opportunities for the 22-25 age bracket, with a 16 percent decline in employment for AI-exposed occupations. This is the old world dying before the new world has been born. 

Other research points towards a shift in requirements for graduates, with an ISE study finding that 58 percent of companies expect some change in tasks and responsibility and a further 29 percent anticipate significant changes. For the generation entering the workforce, education and upskilling will be required to be prepared for the evolving world of employment.

For New Zealand, the emergence and development of AI provide a potentially unique opportunity to deal with one of our persistent issues. We were once on a par with Scandinavian countries in productivity, but we now have 40 percent less output per hour than Denmark, Finland, or Sweden. 

This has been driven by an overreliance on labour force expansion instead of investment in technological innovation. AI is seen as part of the solution by politicians, as it is a potential avenue to address this issue and improve outcomes for New Zealand’s economy.

New Zealand firms have been overwhelmingly embracing AI, with 87 percent using AI to some degree, compared to 75 percent globally. Small-to-medium enterprises (SMEs) have an 82 percent adoption rate, while large enterprises are reaching nearly 92 percent. However, firms are identifying that its effectiveness is being held back by a lack of internal capability and skills, as well as data quality and integration issues.

High adoption without complementary investment in skills and institutional change may deepen disruption rather than resolve it. This tension captures the present moment. Productivity gains are promised, firm and institutional transitions are emerging, and the impact on people is in the balance.