#28 - The Art of Prompt Design, OpenAI Codex, Fine Tuning and More with Author David Shapiro (daveshapautomator)

Multimodal by Bakz T. Future

English - February 10, 2022 04:16 - 1 hour - 45.2 MB - ★★★★ - 7 ratings
Technology Homepage Download Apple Podcasts Google Podcasts Overcast Castro Pocket Casts RSS feed

Previous Episode: #27 - OpenAI’s New Embeddings API Endpoint, InstructGPT Announcement (Fine Tuning Research)

Next Episode: #29 - OpenAI’s InstructGPT is a Game Changer!

Welcome back to Multimodal! Today's very special guest is David Shapiro. He's an author, IT professional, and frequent OpenAI community forums contributor.

In this wide ranging discussion, we discuss the art of prompt design, his thoughts on InstructGPT, OpenAI codex, his new book, the OpenAI community at large, and share thoughts on fine tuning and so much more.

Podcast Discussion Timestamps

00:00 - Intro
02:48 - How David heard about OpenAI's GPT-3
05:57 - David's first time trying GPT-3/his favourite use cases
10:38 - Did GPT-3 feel advanced to you?
15:00 - What are the keys to great GPT-3 prompt design?
18:34 - How do you become a better writer?
25:43 - GPT-3: The end of the socially awkward developer (?)
30:16 - Is GPT-3 Prompt Design Over? (Chatting About InstructGPT)
34:21 - Open Fine Tuning Lessons Learned
42:00 - OpenAI Codex / David's Billion Dollar Use Case Idea
50:00 - Which AI Models David is using currently
53:15 - David's Book - Natural Language Cognitive Architecture
1:01:25 - What is the pathway from now to true AGI?
1:08:41 - Thoughts/feedback about the OpenAI Community Forums
1:18:40 - My Feedback on OpenAI's Community Forums
1:22:00 - My Feedback on OpenAI's Community Forums II #hottake
1:24:46 - Multimodal AI technology
1:27:43 - What would you do with Multimodal AI?
1:32:38 - Where do you see all of this stuff going? 5-10 years
1:35:24 - Closing thoughts