In a fresh development, if you ask them a simple question, you’ll frequently get multi-part answers, complete with bullets and emojis. ChatGPT and other big AI chatbots aren’t ones for holding anything back.
Industry observers note that if your intent is to actually learn something rather than being on the receiving end of a term paper, ChatGPT’s lengthy replies might be little more than a waste of tokens. Those giant AI-generated answers can be overwhelming.
As part of the ongoing story, i didn’t get far before my eyes started to glaze over. For example, when I asked ChatGPT (the “instant” version) to teach me something about neural networks, a key AI concept that I’m still wrapping my head around, it spat out an 800-word answer in 15 parts–no kidding.
As part of the ongoing story, this time, I have a prompt that takes a different approach to learning: making an AI teach you about a topic by asking you questions about it. Now, I’ve already written about an AI prompt that helps you learn 80 percent of a concept by distilling the top 20 percent of what you need to know about it (the “80/20” prompt).
Industry observers note that do not explain it to me yet. Instead, ask me a series of questions—one at a time—to help me construct the understanding myself. Only provide the full explanation once I have correctly identified the core principles through my own reasoning. I want to learn [Concept].
In a fresh development, the Socratic method prompt is a widely used way to turn a lecture-prone AI into a classically-trained guide. Yes, we’re talking about the Socratic method, a teaching method that involves asking a student a series of probing questions rather than spoon-feeding the answers.
As part of the ongoing story, when I appended the Socratic method prompt to my original “teach me about neural networks,” I got this response:.
The report highlights that i won’t explain neural networks yet. We’ll build the idea together, one question at a time. Perfect.
The report highlights that imagine you’re trying to build a computer program that can recognize whether a photo contains a cat.
As part of the ongoing story, what kinds of things do you think the program could look for in the image to help decide “cat” versus “not cat”?
The report highlights that (I’m still struggling through the relevant chapter in Melanie Mitchell’s Artificial Intelligence: A Guide for Thinking Humans, which is required reading for those interested in the history of AI.). In its answer, ChatGPT is kicking off a classic lesson about how neural network-powered AI systems can be trained to recognize images of cats.
As part of the ongoing story, after a few more questions and answers, we went from discussing clues that would distinguish cat photos from other images (“whiskers”) to the concept of weights in AI model training.
In a fresh development, i actually learned something. And you know what?
As part of the ongoing story, his coverage of artificial intelligence interrogates the most recent LLMs, and how they can be used at work and at home to be best prepared for the AI revolution. “AI is going to change our lives sooner than we think,” Ben writes. “Our best way to adapt is by using it every day.” Ben has been a PCWorld author since 2014, and has covered everything from laptops to security cameras before launching PCWorld’s AI beat. Ben's articles have also appeared in PC Magazine, TIME, Wired, CNET, Men's Fitness, Mobile Magazine, and more. Ben holds a master's degree in English literature. Ben has been writing about consumer technology for more than 20 years, and now focuses his reporting on AI as it relates to the basic human experience.