Resources: Projected AI progress

Exercises: Projected AI progress

Resources: Today's AI systems

Exercises: Today's AI systems

Resources: Distribution

Exercises: Distribution

Next steps: Programs

Resources: Automation

Exercises: Automation

Resources: Explosive growth

Exercises: Explosive growth

In this unit, we'll explore the potential for AI to dramatically accelerate economic growth, potentially to explosive levels. We'll examine both theoretical models of growth acceleration and historical precedents, with particular attention to the conditions that enable or prevent such acceleration.
Understanding explosive growth requires careful analysis of multiple feedback loops - from AI automating R&D, to the accumulation of digital workers, to the interaction between population and innovation. We'll examine how these mechanisms might differ from historical patterns of growth acceleration, and what factors might prevent or enable them. Special attention will be paid to scale effects, elasticities of substitution, and the distinction between cognitive and physical automation.
As you study these models, consider how different assumptions about technological capabilities, economic parameters, and human preferences lead to dramatically different conclusions about AI's potential impact on growth. The difference between steady exponential growth and genuine explosive growth often hinges on subtle modeling choices.

By the end of the unit, you should be able to:
\- Understand the conditions for explosive growth
\- Understand potential bottlenecks, and the arguments for and against them mattering
\- Compare with historical growth accelerations and their limitations


Many economists are sceptical that AI will lead to explosive growth. This may partly stem from an underestimation of future technological advancements, but economic mechanisms are likely a significant factor. Many of these might be non-obvious and relate to ongoing research in economics.

**List and rank (on likelihood) as many economic mechanisms as you can imagine that could prevent a significant boost in AI capabilities from contributing to growth. Do any of the papers provide channels by which these problems could be overcome? (15-30 minutes)**

You may want to consider:
- Production networks
    - Elasticities, returns to scale, factor reallocation
- Diffusion
- Income and wealth effects
- Market concentration and markups
- The innovation literature
- Political economy and regulation
- Factor accumulation and substitutability



Economics of TAI

Exercises: Explosive growth

Exercises