Appendix C — Large Language Models (“AI”) ✏️
Coding is an integral part of real-world climate-risk analysis, and large language models (LLMs; often referred to as “AI” models) are rapidly changing how some kinds of coding happen. Beyond web-based chatbots, you may have useed tools like GitHub Copilot (free for students and educators) or Claude Code (see free Deeplearning.AI Course). LLMs use powerful new technologies that can support learning and replace tedious tasks, but they can also threaten your intellectual growth and skill development (Kosmyna et al. 2025; Bastani et al. 2025).
It is clear that there are some tasks that should be delegated to these models and some tasks that must remain human-driven. However, there are tremendous differences of opinion about how most tasks in the middle can or should be allocated. As you wrestle with these questions for yourself, you should explore resources like:
- AI Snake Oil is a blog that seeks to dispel hype, remove misconceptions, and clarify the limits of AI. The authors are in the Princeton University Department of Computer Science.
- AI software assistants make the hardest kinds of bugs to spot from Pluralistic is a thoughtful and deep blog post about the perils of (mis)using LLMs for coding.
- One Useful Thing is a newsletter about AI focused on implications for work and education. The authors’ prompt library is also a good resource for working with LLMs.
- Ed Zitron’s Where’s Your Ed At is a newsletter that takes a critical perspective on the business models and hype narratives around AI.