The release of ChatGPT in November 2022 has raised great expectations that artificial intelligence (AI) can contribute to solving problems in many fields and improving productivity, but it has also raised concerns that many jobs will disappear and income inequality will widen further.
AI is commonly perceived as a general-purpose technology, like the internal combustion engine, early electrical-based technologies, and computers.
These technologies have the potential to disrupt large parts of the economy and put many workers out of work.
Every wave of innovation throughout history has generated fears of “technological unemployment,” but economies have so far generally been successful in creating more jobs than they have lost, despite often painful transitions.
Automation typically leads to increased productivity, which generates income that can be spent on new products and services, which in turn creates demand for labor.
Employment in activities that are hard to automate is growing along with aggregate demand, and new jobs are being created.
David Autor of the Massachusetts Institute of Technology and his co-authors found that roughly 60% of U.S. jobs in 2018 did not exist in 1940.
Labor reallocation due to technological change often poses challenges for workers who need to be retrained and upskilled.
Structural change can also lead to macroeconomic and financial instability if economies struggle to adapt.
Income distribution has been affected, as shown by the polarisation of labour markets in many developed countries over the past few decades.
Will the development of labor markets in the AI ​​era follow historical patterns, or will AI be a game-changer?
Like many digital tools before it, AI has great potential for deployment across a wide range of economic activities.
Until recently, we were primarily able to automate routine tasks.
AI offers the potential to automate non-routine cognitive tasks due to its ability to learn on its own from vast datasets and integrate implicit knowledge.
Some researchers believe that machines will eventually surpass humans in almost every activity, but even if we believe such predictions, that’s probably a long way off.
Currently, AI can automate some cognitive tasks and complement humans in performing more complex tasks.
Several studies have investigated exposure to AI in different occupations by linking some AI applications to the skills used in specific occupations.
This is a good starting point for thinking about the potential impact of AI on employment, productivity, and inequality.
Research from the Organisation for Economic Co-operation and Development’s (OECD) World Forum on Productivity suggests that jobs in knowledge-intensive services sectors such as finance, advertising, consulting and information and communications will be most vulnerable to the impact of AI.
Conversely, workers in mining and construction, or in less knowledge-intensive service sectors such as administrative and support, transport, water and waste management, have limited exposure to AI, raising concerns that following the automation of routine tasks that mainly affect blue-collar workers, technological developments could lead to the automation of knowledge-intensive tasks as well, threatening the employment of highly skilled workers.
Key Questions
A key question is whether AI will primarily automate tasks, or whether it also has the potential to augment workers’ job performance.
In other words, will AI primarily be a substitute for human labor, or a complement?
According to a report from the World Economic Forum, both aspects are important, with their strength varying across economic sectors. Tasks that don’t require creatively solving ambiguous problems, collaborating with others in real time, or validating outputs are considered automatable.
The potential to increase worker capabilities is particularly strong in knowledge-intensive services.
For example, in information technology services, about 32% of the time spent on all tasks could be subject to automation, but the expansion potential is even higher, reaching almost 37% of the time spent on all tasks.
In the media and publishing industry, the potential for augmentation far outweighs the potential for automation, at 32% of time spent on all tasks versus 24%.
So while AI may displace some workers, it will also increase the productivity of others.
Medical Diagnosis
To take a more concrete example, AI can improve medical diagnoses and help structure litigation, but it is unlikely to replace doctors or lawyers anytime soon.
While AI provides skilled professionals with valuable new tools, human intervention is still often required, especially for critical decisions, as AI can hallucinate and produce inaccurate or misleading results.
Recent studies have shown that AI has significantly improved worker performance in certain tasks, ranging from 14% in customer service to 56% in coding.
This suggests that AI could deliver significant productivity gains if it were widely deployed across the economy.
But linking micro-level performance to economy-wide productivity is difficult.
Increased Productivity
Estimates of the potential for AI-driven labor productivity gains in developed countries over the next decade generally vary from about 0.1 percentage points per year to more than 2.5 percentage points.
These figures show that U.S. nonfarm labor productivity will grow at an average annual rate of 2.1% between 1947 and 2023, and 1.5% since 2006.
Thus, according to the most optimistic estimates, AI will boost productivity growth to levels far above historical trends.
The median estimate is that labor productivity growth would approach the 3 percent rate observed between 1996 and 2005, during the period of widespread Internet adoption.
But other studies suggest this may be overly optimistic.
AI-related productivity gains may vary by country.
South Korea is well positioned to be one of the main beneficiaries thanks to its high level of digital development and strong position in the global semiconductor market. — The Korea Herald/ANN
Christophe André is a Senior Economist at the OECD. The views expressed here are his own.
#hurt #jobs #increase #inequality