Better AI models enable more ambitious work
Source: https://cursor.com/blog/better-models-ambitious-work
Date: 2026-04
Authors: Cursor Research + Suproteem Sarkar (University of Chicago Booth)
Key Findings
Jevons Effect in AI
- AI usage (weekly messages per user) increased 44% during 8-month study
- Better AI → greater AI demand (not substitution)
- Efficiency gains increase total consumption, not reduce it
Study Design
- 500 companies using Cursor
- July 2025 through March 2026
- Included Opus 4.5 and GPT-5.2 releases
Sector Differences
| Sector |
Increase |
| Media and advertising |
+54% |
| Software and dev tools |
+47% |
| Finance and fintech |
+45% |
- Finance: arms-race dynamic — competitive pressure to adopt
- Media/advertising: greenfield opportunities expanded by better models
Complexity Shift ("Shift Right")
- Initial: more of same complexity with better models
- After 4-6 week lag: more complex tasks
- Low complexity messages: +22%
- High complexity messages: +68% (mostly in last 6 weeks)
Task Distribution Changes
| Task Category |
Increase |
| Documentation |
+62% |
| Architecture |
+52% |
| Code review |
+51% |
| Learning |
+50% |
| UI/styling |
+15% |
- As AI generates more code: need for documentation, review, understanding grows
- Developer role shifts from coding to managing AI output
- Cross-system tasks (architecture, deployment) increase
Economic Impact
- AI does both: facilitates existing work AND opens new productive opportunities
- Expansion may eventually be the bigger story
- Not just substitution but augmentation and expansion
Key Insight
- Better models don't just make developers faster — they enable fundamentally different, more complex work
- The lag in complexity adoption reflects discovery time + workflow reorientation
- The Jevons effect suggests AI demand will grow with capability, not saturate
Forrás
https://cursor.com/blog/better-models-ambitious-work