NotebookLM Just Killed After Effects? + ChatGPT vs Anthropic War
Summary
Hosts Matt Wolfe and Joe Fear analyze recent AI model releases, including OpenAI’s GPT 5.3 and 5.4, and Google’s Gemini 3.1. They explain training phases like fine-tuning and distillation, highlighting improvements in conversational tone, massive context windows, and native computer use for agents. The episode also covers Notebook LM’s new cinematic video generation capabilities for Ultra plan users, offering insights into how these tools impact everyday users and developers.
Topics discussed
Intro and welcome to the Next Wave Podcast
Overview of the week's AI model releases
Grok 4.20 and OpenAI/Google model updates
ChatGPT 5.3 Instant: tone and preamble changes
Explaining LLM training: pre-training vs fine-tuning
Why recent updates are likely fine-tuning only
GPT 5.2 Mini and the meaning of version numbers
GPT 5.4 features: 1M context window and coding
New agent features: computer use and tool search
Agent mode availability and interactive controls
Gemini 3.1 Flashlight: distillation and speed
Practical use case: YouTube thumbnail analysis app
Google AI Ultra plan and NotebookLM video features
Demo: Generating cinematic videos in NotebookLM
The Pentagon-Anthropic dispute and user migration
Claude's rise in popularity and market share shifts
Ben Affleck's AI filmmaking tool and Netflix deal
Interview: Impact of AI on Hollywood jobs and costs
Meta smart glasses privacy scandal and lawsuits
Outro and call to action
Listen ad-free on Castria