Macro Markets and Machines_The Economic and Market Transformation Driven by AI_GWM report - Flipbook - Page 27
Macro, Markets, and Machines
November 2025
Image created by Scotia Wealth Management.
Section 4.1: AI Productivity Channels
Acemoglu (MIT, 2024) evaluates the possible impact of AI for a representative economy with a simplified, yet
intuitive and academically robust, framework for analyzing the possible benefits and drawbacks of increased AI
adoption.
In his research, he outlines four main channels of impact on productivity: automation, complementarities,
deepening of automation, and new labour-intensive products or tasks. The first two are well known and fairly
intuitive, generally serving as the key starting points for research into the possible impacts of AI, while the latter
two are smaller factors to consider.
Automation. AI widens the use of automation to a greater share of tasks in the production process, fully
replacing human labour input owing to greater efficiency, with the likely benefit of reducing per-unit total costs
to maximize profits. In Acemoglu’s model, greater automation takes place under AI as the latter is introduced as
a relatively lower-cost option for tasks that were not previously fully automated (e.g., basic writing, data
classification). For some firms, its introduction may also enhance the effectiveness of other automated tasks,
thus increasing automation as an input into production at the detriment of human labour.
Complementarities. AI increases the productivity of workers in labour-performed tasks. This could be by
providing workers with better information more quickly or by automating subtasks (e.g., writing basic initial
code). AI usage would allow workers to specialize in other tasks that are not fit for automation, likely increasing
their performance or productivity in a differentiated output. The increase in overall labour productivity from
workers being able to dedicate relatively more time to non-automated subtasks (as others are automated)
would be a key driver for greater labour demand under expanded AI adoption.
Deepening of automation. Tasks that are already automated exhibit greater performance or reduced costs
thanks to new AI technology, generating positive productivity effects by merely replacing or “upgrading”
currently automated processes that had already supplanted labour supply.
New labour-intensive products or tasks. The displacement of labour may create a greater number of workers
who would be free for employment in new activities where labour has an advantage over automation
(reskilling/retraining notwithstanding) or fill employment gaps in certain sectors, lifting overall economic output.
Advances in technology, such as in AI, could also make possible the creation of new activities that generate
demand for labour. New tasks are an important counterweight to automation in maintaining demand for labour
(as has been the experience of previous episodes of automation).
Scotia Wealth Management
26
