---
id: "concept-parenting-ai"
type: "concept"
source_timestamps: ["01:30:50"]
tags: ["prompt-engineering", "mindset"]
related: ["concept-ai-as-guesser", "framework-parenting-ai"]
sources: ["day2"]
sourceVaultSlug: "ai-human-harmony-atoms-agents-2026Apr25"
originDay: 2
---
# Parenting vs. Programming AI

## Parenting vs. Programming AI

The shift in approach required to get high-quality outputs from AI. Because AI is a [[concept-ai-as-guesser|predictive engine]] rather than a deterministic software program, you cannot "program" it with rigid code. Instead, you must "parent" it.

This involves:
- **Providing deep context** about who you are, what your business does, and what you care about
- **Giving clear examples** of what good output looks like
- **Offering continuous, iterative feedback** when it makes mistakes

Much like teaching a child or onboarding a new employee, the relationship with your AI improves proportionally to the quality and volume of context you invest.

The practical methodology is codified in the [[framework-parenting-ai|Parenting AI Framework]]. As [[entity-lior-weinstein|Lior Weinstein]] puts it: [[quote-parent-ai|"Don't program it. Parent it."]]

> **Enrichment note:** This approach is validated by prompt engineering research (e.g., Lil'Log's Prompt Engineering Guide, 2023 updates), which shows that chain-of-thought prompting, few-shot examples, and feedback loops yield 20–50% quality improvements on benchmarks.

**Definition:** The methodology of guiding AI through context, examples, and iterative feedback rather than rigid, deterministic coding.

---
*See also: [[action-tone-mirror]], [[contrarian-stop-chasing-tools]], [[concept-ai-clone-first-hire]]*


## Related across days
- [[framework-context-sandwich-d1]]
- [[framework-context-sandwich-d3]]
- [[concept-playbooking-method]]
- [[framework-playbook-outline]]