**Generative AI Unschool** --- **Today's Journey** - Principles - Interactive demos - Putting it into action - Q&A --- **Why**  --- > This is the most fun I've had coding in years -- Dan Sofer --- **What**  --- **Principles** - There's no AI instruction manual - Coding with LLMs is an iterative process - To get the best results think: "open book test" --- **No user guide** *Proficiency in working with AI comes through experimentation and experience*  https://www.oneusefulthing.org/p/centaurs-and-cyborgs-on-the-jagged --- **Coding with AI should feel like a conversation** - Think of it as a dialogue with an ever-enthusiastic but error prone intern --- - Engage in a dialogue with the AI: ask it questions, give it feedback, and challenge it to do better - https://chat.openai.com/share/5fce83eb-6f72-4e91-84a1-b2aeb61ecc82 --- **ChatGPT is not a knowledge database, it's a resoning engine** - in other words, **it's not google**, and you won't get good results if that's your mental model for it - data cuts off in sept. 2021, so it won't have access to latest API docs --- **Giving AI Open Book Tests** - Provide ample context - Give it a way to answer your question as well as the question - https://every.to/chain-of-thought/gpt-4-is-a-reasoning-engine --- - **Example of open book test**: scrutinising potential purchases --- **Exercise: ChatGPT & LinkedIn** - Ask ChatGPT who you are and assess how good the results are - Now copy and paste your LinkedIn details and ask it who you are again --- Write a "Hello World" Python Program in Replit --- Write a file! (then read it) --- Write a web scraper --- Build a simple chatbot --- **Wrapping Up** - Questions? - Thank you for participating!