Want Can Should May Framework

The want-can-should-may framework is Wei Qing / 韦青’s engineering judgment frame in E42 孟岩对话韦青：沉默的主角. It asks whether people want to do something, can do it, should do it, and may or are ready to do it. The point is to make technical capability only one part of the decision rather than the whole decision.

The framework fits the episode’s broader warning that AI-era engineering should not collapse into capability worship. A system can be buildable and still be undesirable, premature, harmful, socially unready, or misaligned with the human state of the people using it.

Key Claims

• “Can” is necessary but insufficient for engineering judgment.
• “Want” connects technical work to human desire and purpose.
• “Should” forces ethical, social, and organizational consequences into the decision.
• “May” adds permission, readiness, constraint, and context.
• In AI work, the framework is a guardrail against treating every model capability as an automatic product direction.

Connections

• Wei Qing / 韦青 — source speaker.
• Human Agency Under AI — choosing what deserves delegation or building.
• Human Judgment Under AI — deciding whether an output or technical path fits the situation.
• AI Engineering Thinking — practical translation of goals, constraints, tests, and workflows into AI-assisted work.
• AI Governance And Compliance — adjacent governance layer for what AI systems are allowed to do.