Catch up

 大型語言模型 就像人生出來直接讀博班，嬰兒會的東西反而不會。嬰兒知道洋娃娃被蓋住了並沒有不見，但AI 會以為憑空消失了。

來自ICML2024的論文

https://www.facebook.com/share/p/1EMzRKUrBQ/?mibextid=wwXIfr

How many r in strawberry

How many characters in "sentence"

LLM 把人類知識灌頂給 AI

GPT, Generative, Pre-training, Transformer

如何取出知識

endgame

化繁為簡 思維鏈 (chain of thought)

深謀遠慮 RL(reinforced learning) What is Reinforcement

AlphaGo AlphaZero

pre-training vs. learning

nature vs nurture aspect of snake fear

AGI?

AI 可以靠自己學習，探索，發現新事物嗎？

Follow up to unit 3-4

 Hide panel unless requested (see next)

Voronoi vs non Voronoi navigator 

Repeated failures, how to overcome 

• repeated tries, new chat rural road
• repeated tries, new chat child safety

Error running artifact

An error occurred while trying to run the generated artifact.

• Unexpected token, expected "," (1:11)

Unit 5: 環境安全檢測與評估視覺化

Hazard detection (環境危害偵測)

意外事件預測與模擬

Home Hazard chain identification and animation （居家危害鏈偵測與危害視覺化）

CL#3 Obstacle detection

  建議工具

使用 Claude Sonnet 4 or 4.5

使用 ChatGPT 5

使用 Gemini 2.5 Pro  or Gemini 3 Pro 免費額度最高 1M tokens

使用 Grok 4

 課堂練習 

Deadline: Saturday at 23:59

Send all the share links to  me chang212@gmail.com by email with subject CL#3  [your id, your name]

How to share a Claude link

How to share a ChatGPT link

How to share a Grok link

How to share Gemini Link

至少做1題

1. Referring to Global Obstacle detection, implement it using Photo gallery or your image banks. 

2. Referring to Dynamic Obstacle detection, implement it using Photo gallery or your image banks. 

3. Referring to Simulating a navigation assistant, implement it using Photo gallery or your image banks. 

4. Goggle maps for People with BLV (Blindness and low vision) urban

https://claude.ai/public/artifacts/a8cebec4-38c0-4464-803a-e53abf856cc1 (default)

https://claude.ai/share/ec38bb69-c956-455d-8720-62d5579eddea

https://claude.ai/public/artifacts/f692fa50-ee15-42a9-8dc4-b3e982469dc6 (A*)

Photo gallery

5. 製作3d 障礙模型圖，模擬行進

Your prompts show a clear progression from initial obstacle identification → edge detection → 3D visualization → realistic models → collision detection → intelligent avoidance → critical safety zones.

ＡＩ障礙偵測，行進模擬 (artifact)

Catch up

撰寫AI App (Sonnet 4.5 new)

繪圖 (Sonnet 4.5 enhanced)

 AI 時代找文獻的方式 

ＡＩ時代是人類在做研究，還是ＡＩ在做研究 (你行你來)

烏克蘭防空模擬

使用合併模組的方式製作複雜場景的空防系統模擬（使用電腦大螢幕）- 過程解說

3d障礙分析(模擬行進)

 

製作3d 障礙模型圖，模擬盲人行進 

Your prompts show a clear progression from initial obstacle identification → edge detection → 3D visualization → realistic models → collision detection → intelligent avoidance → critical safety zones.

ＡＩ障礙偵測，行進模擬 (artifact)