Jim talks with Dan Shipper about practical uses of GPT-3 and ChatGPT at the personal scale. They discuss how Dan started playing with these tools, the feeling of new generative AIs, GPT-3 vs ChatGPT, writing a screenplay using ChatGPT, using GPT-3 to analyze journal entries, circumventing the context window limitation, GPT-3 as a journaling tool, how ChatGPT does embedding, the coming market for chatbot personas, the value of guardrails, the monetary cost of using GPT-3, solving the organizational problems of note-taking, Stephen Reid’s knowledge-graph of this podcast, the invention of the graphic web browser & the frozen accidents of HTTP & HTML, meta-prompts & data pipelines, how Yohei Nakajima eliminates repetitive tasks using LLMs, and much more.
- Episode Transcript
- Chain of Thought (Every)
- “Can GPT-3 Explain My Past and Tell My Future?”, by Dan Shipper
- GPT Index
- “Chat GPT ‘DAN’ (and other ‘Jailbreaks’)”
- JRS Knowledgegraph, by Stephen Reid