Peter Berezin: AI Boom's Economic Impact is Less Than Believed

AI and Productivity Data

• 💡 Aggregate productivity data has not yet shown a significant impact from generative AI, growing at a pace consistent with prior projections.
• ⏳ Similar to the adoption of PCs and the World Wide Web, AI's broad-based positive impact on productivity may take considerable time to materialize.
• 📉 Historically, even after widespread computer adoption, productivity growth only accelerated in the mid-1990s, suggesting a similar lag for AI.

Investor Patience and Profitability

• ⚠️ Investors may not have the patience to wait for AI-driven productivity gains to translate into increased profits.
• 📈 During the dot-com era, productivity rose around 1995, but profits from the internet didn't appear until 2005, highlighting a potential delay between productivity and profits.

Network Effects and AI Business Models

• 🚫 Unlike software benefiting from network effects (e.g., Office, Instagram), AI interactions are primarily with the AI itself, lacking this advantage.
• 💸 AI business models face challenges due to high capital expenditure requirements for data centers and chips, similar to the airline industry where products are commoditized.
• 🧩 The similarity in AI architectures and high infrastructure costs suggest that AI may not inherently lead to widespread profitability across industries, despite its importance.