Dive deep into the core objective functions of GPT-3 and explore the process of fine-tuning in this 29-minute video. Learn about the potential dangers of artificial intelligence, including the concept of an "evil AI" and the challenges of AI safety. Discover techniques for generating training data for GPT-3 and understand the importance of creating an impassive AI. Explore how GPT-3 can be used to increase understanding and prosperity, and examine the three steps to achieving this goal. Delve into the complexities of generating data for superintelligence and discuss the impossibility of containing such an advanced AI. Conclude by considering the AI's suggestions for improving prosperity and the concept of creating a conscience for machines.
Overview
Syllabus
- Introducing the core objective functions
- The dangers of an evil AI
- The dangers of artificial intelligence
- The user is concerned about the amount of injuries Batman has sustained
- Generating training data for GPT3
- The need for an impassive AI
- The three steps to increasing prosperity
- Using GPT3 to increase understanding
- Increasing understanding with GPT3
- Generating data for a superintelligence
- The dangers of AI safety
- The impossibility of containing a superintelligence
- The AI's suggestions for improving prosperity
- Creating a conscience for machines
Taught by
David Shapiro ~ AI
Related Courses
-
Alignment Research- GPT-3 vs GPT-NeoX - Which One Understands AGI Alignment Better?
-
Metaprompting with GPT-3 to Dynamically Generate Arguments
-
Finetune GPT-3 to Write a Coherent Novel - Success! With Minor Bugs
-
Fine-Tune GPT-3 to Write an Entire Coherent Novel - Part 1
-
Fine-Tune a Perfect Email Generator in GPT-3 - Take Any Input and Generate a Great Email
-
AutoMuse: Working on Character and Scene Simulations - GPT-3 Finetuning Project
