Context Engineering vs. Prompt Engineering: - AI動画分析

AIコメンタリー

動画を再生してAIコメンタリーを見る

Okay, so prompt engineering is what we're all used to, crafting the text to get the AI to do what we want. But this idea of context engineering, building out *everything* the AI sees, that's a whole new level.
Ah, Secret Agent Graeme, the travel expert! This is a great way to visualize the problem – a simple request, and bam, you get a hotel in the wrong Paris. It's relatable because it highlights how easy it is for these systems to misunderstand without enough specific guidance.
Ha, Paris, Kentucky! That's a classic example of context being king. It shows that just having the right words isn't enough; the AI needs the *world* around the request to be understood. This really makes you think about how much information is implicitly assumed in our requests.

もっと見たいですか？サインアップして全ての会話を見る

新規登録

動画の要約は視聴を開始すると表示されます

Prompt engineering, the craft of designing input text for large language models (LLMs) [0:00], focuses on steering the model's output through instructions and examples. Context engineering, however, is a broader system-level discipline that programmatically assembles *all* information an LLM encounters during inference [0:23]. This includes not just prompts but also retrieved documents, memory, and tools. A travel booking agent example illustrates the difference: initially booking a hotel in Paris, Kentucky, instead of Paris, France, due to insufficient context [0:47]. This highlights that while prompt engineering addresses the wording of the request, context engineering ensures the LLM has the necessary environmental information, like the conference location, to plausibly complete the...
全機能を利用するには

サインアップまたはログインして、完全な動画分析機能にアクセスしましょう

新規登録 ログイン

現在のセクション要約

動画の要約は視聴を開始すると表示されます

Prompt engineering, the craft of designing input text for large language models (LLMs) [0:00], focuses on steering the model's output through instructions and examples. Context engineering, however, is a broader system-level discipline that programmatically assembles *all* information an LLM encounters during inference [0:23]. This includes not just prompts but also retrieved documents, memory, and tools. A travel booking agent example illustrates the difference: initially booking a hotel in Paris, Kentucky, instead of Paris, France, due to insufficient context [0:47]. This highlights that while prompt engineering addresses the wording of the request, context engineering ensures the LLM has the necessary environmental information, like the conference location, to plausibly complete the...
全機能を利用するには

サインアップまたはログインして、完全な動画分析機能にアクセスしましょう

新規登録 ログイン