Software engineers will become scientists: Managing AI agents and leveraging non-digital context
gajesh · x · 2026-08-24
Alex Attalah's article posits that software engineers must evolve into scientists to remain valuable in the AI age, specifically by managing an army of AI agents. High-leverage management involves either great taste (intuition-based judgment) or great science (data-based prediction).
Scientific Thinking for Engineers:
- Establish target metrics (product or business KPIs)
- Prepare test and validation data
- A/B test new prompts and techniques
- Consider edge cases and pull in context AI lacks
- Measure results
The Human Advantage: Humans hold a structural advantage in acquiring context from non-digital sources (e.g., the physical world). This scarce context becomes extremely powerful when applied intelligently, potentially reshaping organizational investigations. While AI will eventually conduct its own science, humans who explore new data sources independently will retain a slight edge.
More from coding & agent
- Ask HN: Best way to add vision support to Deepseek v4 Flash on DGX? — StartupTim · 2026-08-24
- Context Layer: A User-Controlled Contract for Cross-Model/Agent Context Transfer — sierracatalina · 2026-08-24
- Gemini CLI Fix: Prevents Output Doubling on Invalid Truncation — Eswar809 · 2026-08-24
- Test: ChatGPT Desktop 5.6 Sol Outperforms Cursor + Grok — SimonBalmain · 2026-08-24
- Hugging Face Runs an Overnight Agent to Rescue Open-Source Models Buried on arXiv — NielsRogge · 2026-08-24
- Dev asks: Did Cursor actually bring people back to IDEs? — ChrisRackauckas · 2026-08-24