The next innovation revolution - powered by AI
Date read: 12th November 2025
Paper link
Key Points
- Previous revolution example: Industrial Revolution
- World GDP per capita stayed roughly constant from 0 to 1750
- It has 14x since then. Life expectancy has also shot up
- All these improvements were built on top of scientific innovations. E.g. steam engine, vaccines
- Modern issue: research and development is getting more expensive and less productive
- Maintaining Moore’s Law is becoming more expensive
- Eroom’s Law: Number of novel drugs produced per billion dollars spent has halved every 9 years from 1951 to 2011.
- How can AI help tackle this?
- Increase velocity, volume and variety of design generation
- Generate more designs quickly with more variety
- Real world example: store designs, protein design
- Accelerate evolution of candidates
- Approximate things like FEA and physical tests using surrogates
- real world examples: wether prediction
- Accelerate research operations
- Ideas, code, evaluating customer needs can all be accelerated by LLMs
- Examples:
- Summarising and exploring existing research
- Knowledge management
- Automating documentation
- Idea creation
- Will have most impact in software as it is text based