Environmental outcomes of artificial intelligence

Area of Science:

• Environmental economics
• Technological change
• Climate policy

Background:

• Growing concerns about the environmental impact of technological advancements.
• The need to understand the macroenvironmental consequences of Artificial Intelligence (AI) adoption.
• Assessing AI's role in decarbonization and climate resilience.

Purpose of the Study:

• To investigate the relationship between AI adoption and CO2 emissions (territorial and consumption-based) in advanced economies.
• To evaluate the impact of AI on climate adaptation capacity.
• To analyze the dynamic effects of AI growth shocks on environmental outcomes.

Main Methods:

• Panel data analysis of 30 advanced economies (1995-2020).
• Fixed-effects models with Driscoll-Kraay standard errors.
• Instrumental variable (IV-2SLS) design using geography-filtered patenting shocks.
• Panel local projections for dynamic response analysis.

Main Results:

• Higher AI stock is associated with lower emissions, with evidence supporting emissions reduction in the medium run.
• Territorial emissions initially rise post-AI shock, then decline; consumption-based emissions show persistent decline.
• AI adoption leads to steady improvements in climate readiness (ND-GAIN index).

Conclusions:

• AI has the potential to support decarbonization and enhance climate resilience in advanced economies.
• The environmental payoff of AI is contingent on its deployment pathway and the energy system's capacity.
• Policy interventions may be needed to maximize AI's positive environmental externalities.