Toggle Mobile Menu

Anton Korinek on AI Scaling Laws and Market Structure

Introductions by Markus Brunnermeier
November 26, 2024
12:30 pm
Markus' Academy

More from this series
Zoom

Video & Timestamps

On Tuesday, November 26, Anton Korinek joined Markus’ Academy for a conversation on AI Scaling Laws and Market Structure. Anton Korinek is a Professor of Economics at the University of Virginia and a Visiting Fellow at Brookings.

Watch the full presentation below. You can watch all Markus’ Academy webinars on the Princeton BCF YouTube channel.

Timestamps:

[9:47] The AI Scaling laws

[27:48] Snapshot of the AI market

[39:43] Where is this all going?

[47:54] Concentration concerns

[1:04:58] Regulation and conclusion

Summary

  • A summary in three bullets
    • The effective computing power employed to train AI models is growing by a factor of 10x per year. Investing has been enabled by AI’s scaling laws, whereby performance improves predictably from additional computing power
    • There are concerns that AI models are hitting a “data wall,” but further model improvements could come from synthetic data, more efficient training algorithms, and scaling outside of the training stage
    • The LLM market is highly competitive, but high training costs, network effects, and vertical integration between LLM developers and chips manufacturers raise concerns about future market concentration

Click here for the full summary.