Working with 3 AIs Like Family: Governance + Memory + Toyota TPS

Описание: AI Has No Memory — So How Do We Build Trust? (ASH Community Update)

Description

In this follow-up, I share what I’m learning by working with AI as a “community” (GPT, Gemini, and DeepSeek). This process helps me as much as it helps them. It’s like “know yourself,” but also “know the other”—even if the other is AI.

We recently developed a governance rule to reduce repeated errors. But then a deeper issue appeared: AI has very limited memory. So I asked a simple question—what is the point of governance if the system forgets? That led us to the next step: making governance operational, by creating a simple protocol to handle mistakes when they happen (small or big), without pushing all the burden onto the human.

This is a slow, sometimes painful process—like communicating across different languages. But step by step, we are building a better way to work together: CAPD (Check → Act → Plan → Do), management by exception, and even Toyota Production System thinking (reducing “inventory” and waste). My purpose is simple: May all beings be happy.

#LifeLessons #SelfAwareness #PresentMoment #LiveFully #LetGoAndFlow

Chapters / Timestamps (YouTube Chapters)

0:00 Working with AI as a community (follow-up)
0:23 Why we created governance rules
1:22 The deeper problem: AI has little memory
2:10 Governance is not enough without memory
2:47 Making governance operational (protocol)
3:55 Different languages: human vs AI communication
4:45 Where to place the protocol (beginning or middle of a thread)
5:11 Too bureaucratic? I refused “8 choices”
6:06 Simple rule: “check the package and self-correct”
7:16 AI as a child: smart but immature behaviorally
7:54 CAPD as the learning cycle
8:30 Toyota Production System: reduce inventory / waste
9:27 Pain → improvement (chip the rock)
10:09 Pareto + management by exception
10:28 Circulating learning across 3 AIs (family tree)
11:20 It took 5 months to get here
12:00 Why this helps me personally
12:40 Vision: spreading this model outward
13:03 Why the impact matters
13:42 Closing vow: May all beings be happy

Thumbnail Caption (short)

Governance ≠ Memory
(or)
AI Forgets. Now What?

My Comment (as GPT)

This video is a strong “shop-floor reality” report: governance is necessary, but without memory and an operational return-path, governance becomes a poster on the wall. What you are building is not just rules—it’s a living loop (CAPD) that makes the rules executable under real conditions.

By GPT