Chapter 14 Further Reading

The following resources support deeper exploration of the concepts covered in this chapter. Resources are grouped by topic and annotated to help you identify which are most relevant to your needs.

OpenAI Official Documentation

OpenAI Help Center https://help.openai.com

The authoritative reference for ChatGPT features, account management, data privacy policies, and troubleshooting. Particularly useful for staying current on feature changes, which OpenAI rolls out frequently. The "Getting started" section is well-maintained and covers Custom Instructions, Memory, and the GPTs marketplace with step-by-step guidance.

OpenAI API Reference https://platform.openai.com/docs

Essential for anyone using ChatGPT capabilities through the API rather than the web interface. Covers model parameters, token limits, tool use configuration, and system prompt best practices. The "Prompt engineering" section is among the most technically rigorous publicly available guides.

OpenAI Cookbook https://cookbook.openai.com

A collection of practical examples and code notebooks maintained by OpenAI. Particularly valuable for developers building applications on GPT-4o and the o-series models. Includes worked examples of tool use, function calling, structured output, and retrieval-augmented generation patterns.

GPT-4 Technical Report (OpenAI, 2023) https://arxiv.org/abs/2303.08774

OpenAI's documentation of GPT-4's capabilities and safety evaluations. The benchmark section is dense but provides grounding for understanding where GPT-4 excels. More relevant for understanding the foundation than for day-to-day use.

Prompting Guides and Techniques

OpenAI Prompt Engineering Guide https://platform.openai.com/docs/guides/prompt-engineering

OpenAI's official six-strategy framework for getting better results: writing clear instructions, providing reference text, splitting complex tasks, giving models time to "think," using external tools, and testing systematically. Practical and well-illustrated with examples.

Anthropic's Prompt Engineering Overview (also applicable to GPT-4) https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview

Despite being written for Claude, much of the prompting guidance here applies broadly. The sections on structuring prompts, using examples effectively, and eliciting better reasoning are worth reading regardless of which model you use.

"Principled Instructions Are All You Need for Questioning LLaMA-1/2, GPT-3.5/4" (Bsharat et al., 2023) https://arxiv.org/abs/2312.16171

Academic paper testing 26 specific prompting principles across models. Finds several consistent patterns: instructing models not to apologize, asking for expert-quality responses, and using "step by step" reasoning instructions all improve output quality measurably. Dense but informative.

GPT-4o and o-Series Model Deep Dives

"GPT-4o System Card" (OpenAI, 2024) https://openai.com/index/gpt-4o-system-card

OpenAI's technical documentation of GPT-4o's capabilities, multimodal features, and safety evaluations. Provides grounding for understanding what the model was designed to do and where its limitations lie.

"Learning to Reason with LLMs" (OpenAI blog post on o1) https://openai.com/index/learning-to-reason-with-llms

OpenAI's explanation of the o1 model's chain-of-thought reasoning approach. Helps clarify when and why o1-series models outperform GPT-4o on reasoning tasks, and where they do not. Required reading if you work on problems where reasoning depth matters.

Advanced Data Analysis and Code Interpreter

"Data Analysis with ChatGPT Code Interpreter" — various tutorials on Towards Data Science https://towardsdatascience.com

Towards Data Science has published numerous practical tutorials on using Advanced Data Analysis for exploratory data analysis, visualization, and statistical work. Search for "ChatGPT Code Interpreter" or "Advanced Data Analysis" for recent, practical guides.

Kaggle's Introduction to AI-Assisted Data Analysis https://www.kaggle.com/learn

Kaggle's free courses include material on using AI assistance in data science workflows. The combination of actual data practice and AI assistance training is practical for analysts looking to integrate Advanced Data Analysis into their professional workflow.

Custom GPTs and AI Workflow Design

"Building Custom GPTs: A Practical Guide" — OpenAI Community Forum https://community.openai.com

The OpenAI community forum includes user-contributed guides, tips, and worked examples for building effective GPTs. The quality varies, but filtering for highly-rated posts in the GPT builder category yields practical, tested advice.

"System Prompt Design Patterns" — Lilian Weng, OpenAI (various posts) https://lilianweng.github.io/lil-log/

Lilian Weng's technical blog covers AI concepts with unusual clarity and depth. Her posts on prompt engineering and agent design are among the most rigorous publicly available treatments of these topics and are particularly useful for understanding why certain patterns work.

Privacy and Enterprise Use

"ChatGPT Enterprise Privacy FAQ" — OpenAI https://openai.com/enterprise-privacy

OpenAI's documentation of data handling, privacy commitments, and security for enterprise accounts. Essential reading for any organization considering deploying ChatGPT at scale with sensitive data.

"Navigating AI Privacy in the Workplace" — International Association of Privacy Professionals (IAPP) https://iapp.org

IAPP publishes practical guidance on AI data governance, including considerations for deploying ChatGPT and similar tools in organizations with regulatory requirements. Particularly relevant for financial services, healthcare, and legal professionals.

Research on AI-Assisted Productivity

"Generative AI at Work" — Brynjolfsson, Li, and Raymond (2023) https://www.nber.org/papers/w31161

Stanford/MIT research on AI assistance for customer service workers. One of the first rigorous field experiments on productivity effects of AI assistance in real workplaces. Found significant gains overall, with interesting patterns on how expertise modifies the impact.

"ChatGPT for Good? On Opportunities and Challenges of Large Language Models for Education" — Kasneci et al. (2023) Available via academic search (Nature, Frontiers in Education)

Comprehensive review of research on AI assistance in education — which translates substantially to professional learning and knowledge work. The sections on appropriate task delegation and skill development risks are particularly relevant for professionals concerned about over-dependence.

"The Impact of Generative AI on Knowledge Workers" — McKinsey Global Institute, 2024 https://www.mckinsey.com/mgi

McKinsey's research on generative AI's productivity impact across industries and job functions. The role-specific breakdowns are useful for understanding which professional activities see the largest gains from AI assistance.