Economic Impact of AI: Labor
Exploring how AI will reshape the workforce through augmentation, automation, and entrepreneurship
This post continues a previous post in which I started exploring how AI will transform our lives. I painted an optimistic picture: AI could help us outsource routine tasks, allowing us to climb up Maslow’s hierarchy toward self-actualization. I also argued that this future is not guaranteed. Achieving it will require — even distribution of value created by AI, designing better human-AI collaboration, implementing effective regulation, rethinking education, and more. Over the next few months, I will explore these themes, starting with the economics of AI.
AI will affect consumers, firms, and workers. This post focuses on labor. An excellent deep discussion of the impact of AI on labor is in Acemoglu and Restrepo 2018.
Labor Implications of AI
There are three forces are at play: (i) labor augmentation, (ii) automation, and (iii) the creation of new uses for labor.
Labor Augmentation
AI can act as a copilot, enhancing worker productivity by taking over repetitive tasks and expanding human capabilities. For instance, a film editor using AI-based editing tools can edit projects faster and better. As productivity rises, wages often follow. Additionally, when the cost of automated tasks decreases, demand for those services increases, which can drive up demand for complementary non-automated tasks.
This dynamic may seem counterintuitive, but history provides many examples. Bessen (2016) highlighted how ATMs, despite automating cash dispensing, led to an increase in bank teller employment over the next few decades. ATMs reduced the cost of running bank branches, enabling banks to open more locations, which in turn created demand for tellers to handle tasks not fully automated, like customer service and account advising. Similarly, the automation of weaving in the past boosted demand for related tasks, such as spinning, that hadn’t yet been automated.
Labor Displacement
The second force is automation-driven labor displacement. The equation is straightforward: if AI fully automates work, the demand for human labor decreases in those areas.
The Shift to Capital
The interplay of augmentation and automation has a significant economic consequence: a shift in surplus from labor to capital. So we should expect that labor’s share of national income will decrease. This risks deepening wealth inequality, concentrating economic power in the hands of those who own and control AI. Balancing technological progress with human welfare will be one of the defining economic challenges of the AI era.
Why Entrepreneurship Will Be More Important with AI
If the impact of automation outweighs that of augmentation, AI will lead to more job displacement than job enhancement. In such a scenario, the most important counterbalancing force will be the creation of new uses for labor—entrepreneurship. Historically, major waves of technological innovation has spurred entirely new industries and job categories.
Personally, I believe in human ingenuity. In the nineteenth century, people could not have imagined jobs in modern finance or software engineering. Yet, we got there. Similarly, we will come up with new jobs and uses of human mind. But that doesn’t mean the process will be painless. Unlike past technological shifts that unfolded over decades, AI’s rapid advancement compresses these transitions into just a few years. The need of the hour is reskilling — on a timeline that matches the exponential pace of AI’s development. Unfortunately, innovation in reskilling is not keeping pace with AI development today.
Policy Priorities for the AI Era
To navigate these challenges and ensure AI’s benefits are widely shared, labor policy must focus on three key areas:
Reskilling: Equip workers with the skills needed for emerging industries.
Support for Entrepreneurship: Provide incentives and support for new job creation.
Regulation and Taxation: Rebalance the capital-labor dynamic, ensuring value generated by AI is extracted more evenly.
In the next post, I’ll explore how AI will affect consumers and what it means for society at large.
References
D. Acemoglu and P. Restrepo. 2018. Artificial Intelligence, Automation, and Work. NBER working paper. Link here.
J. Bessen. 2016. Learning by Doing: The Real Connection Between Innovation, Wages, and Wealth. New Haven: Yale University Press.
Kartik, clear eyed insight as always!
My 2 bits.
Regulation and taxation: the international “rules” that enable innovation in AI are skewed too. While talent, in significant numbers will be extracted at low cost from the developing world (ex India), accrued value and productivity gains may not be shared with those nations. Clearly IP rights and tech transfer costs hitherto is an example.
In short equitable returns on talent migration must be reckoned and these nations must be at the table in deciding them. Most Nobel laureates model national economic wealth generation not globally shared prosperity- it remains trickle down.
AI, imho must be differently modelled internationally.
Great read sir. AI can augment and automate jobs but it also needs its creators, maintainers and optimisers. While AI can automate many jobs, it also enables people to solve their problems and further develop their services i.e. companies with their solutions, but monopoly of AI can ruin innovation here. I think AI is a super power and it mustn't be with a selected few.