Getting better at LLMs, with Zvi Mowshowitz

In this episode of Complex Systems, Patrick McKenzie (patio11) and Zvi Mowshowitz discuss how to effectively use Large Language Models (LLMs) for various tasks. They delve into the importance of system prompts for customizing LLM behavior, sharing personal examples of prompts they use. The conversation covers strategies for refining LLM outputs, managing memory features, and using LLMs recursively. They also explore the concept of Generative Engine Optimization (GEO) and the balance between paywalls and open internet accessibility for writers. Additionally, they discuss using LLMs for artistic upskilling, critiquing writing, and even as a sounding board for tone and potential implications in text, emphasizing the importance of experimentation and understanding the limitations of LLMs.

Outlines

Part 1: LLM Basics and Personalization

Part 2: Content, Multimodality, and Hallucinations

Part 3: Writing Refinement and Human Interaction

Sign in to continue reading, translating and more.

Continue

Complex Systems with Patrick McKenzie (patio11)

Part 1: LLM Basics and Personalization

Introduction to System Prompts and Customizing LLMs

Personalizing LLM Interactions and Memory Management

Managing LLM Memory and Introduction to Generative Engine Optimization (GEO)

Part 2: Content, Multimodality, and Hallucinations

Content Strategy for LLM Training and Data Ownership

Multimodality and Recursive Use of AIs

Refining System Prompts and Addressing LLM Hallucinations

Iterative Critique and Experimentation with LLMs

Part 3: Writing Refinement and Human Interaction

Avoiding Leading Questions and Leveraging LLMs for Tone Analysis

Refining Writing with LLM Feedback and Maintaining Original Voice

LLMs as Rubber Ducks and Optimizing Interactions

Prompting Humans and the Importance of Attention

Getting better at LLMs, with Zvi Mowshowitz

Complex Systems with Patrick McKenzie (patio11)

Part 1: LLM Basics and Personalization

00:01Introduction to System Prompts and Customizing LLMs

Introduction to System Prompts and Customizing LLMs

04:24Personalizing LLM Interactions and Memory Management

Personalizing LLM Interactions and Memory Management

07:55Managing LLM Memory and Introduction to Generative Engine Optimization (GEO)

Managing LLM Memory and Introduction to Generative Engine Optimization (GEO)

Part 2: Content, Multimodality, and Hallucinations

12:17Content Strategy for LLM Training and Data Ownership

Content Strategy for LLM Training and Data Ownership

17:17Multimodality and Recursive Use of AIs

Multimodality and Recursive Use of AIs

22:26Refining System Prompts and Addressing LLM Hallucinations

Refining System Prompts and Addressing LLM Hallucinations

27:05Iterative Critique and Experimentation with LLMs

Iterative Critique and Experimentation with LLMs

Part 3: Writing Refinement and Human Interaction

31:38Avoiding Leading Questions and Leveraging LLMs for Tone Analysis

Avoiding Leading Questions and Leveraging LLMs for Tone Analysis

36:29Refining Writing with LLM Feedback and Maintaining Original Voice

Refining Writing with LLM Feedback and Maintaining Original Voice

41:14LLMs as Rubber Ducks and Optimizing Interactions

LLMs as Rubber Ducks and Optimizing Interactions

44:45Prompting Humans and the Importance of Attention

Prompting Humans and the Importance of Attention