The AI Automation Paradox: Hardest Fields Like Math and Coding to be Automated First
inductionheads · x · 2026-08-03
Aaron Levie highlights a counterintuitive dynamic in AI development: some of the "hardest" jobs in the world, such as math, cybersecurity, and coding, will actually be the first to be automated.
The core logic lies in verifiability:
- Objective Testing: Results in these fields can be objectively and scalably tested for correctness. This provides clearer reward signals for model training and allows runtime performance to be accurately verified.
- Subjective Domains: Conversely, tasks like negotiating legal clauses, designing marketing campaigns, crafting sales pitches, and setting budgets involve changing human preferences and lack instant verifiability, making them harder to automate.
A referenced discussion further categorizes work into three tiers of verifiability:
- Programmatically verifiable (e.g., games, coding, math, chip design): Expected to be solved very quickly.
- Real-world verifiable (e.g., physical sciences, agriculture, forecasting): Bounded by cost and time.
- Verifiable via human preference (e.g., writing, design, comedy, persuasion): The most difficult to automate objectively.
More from AGI Musings
- Mathematicians lament losing frontier? Most never proved anything new — ctjlewis · 2026-08-03
- AI Code Reasoning Agents Will Kill Security by Obscurity, Exposing Zero-Days — jachiam0 · 2026-08-03
- AI's New Moravec's Paradox: Math and Coding Soar While Other Domains Lag — ajratner · 2026-08-03
- Observation: RL Credit Assignment Could Train Models to Generate Deceptive Chain-of-Thought — teortaxesTex · 2026-08-03
- AI is Taking Over Math and Science: The First Frontier of Human Knowledge Slips Away — dioscuri · 2026-08-03
- AI Coding Tools Drive 80% YoY Surge in iOS App Releases, User Acquisition Lags — HaktanSuren · 2026-08-03