EP31: Fine-Tuned MrBeast Model Results, Chris Makes a Game, AGI Safety Paper + ERNIE GPT

Anthropic and OpenAI continue their awkward dance as they both court developers, while Apple spends millions training the next Siri. And when an AI generates its own MrBeast video, hilarity and fake deaths ensue. Tune in to hear the bros' spicy takes on the latest in the AI world! Consider liking and subbing if you like the show. Thanks for watching! CHAPTERS ====== 00:00 - Cold open: sparks of AGI 00:20 - Fine Tuning a MRBEAST AI Model Experiment Results & Prompt2Model 09:41 - The Realities of AI Theory Vs Reality: Trying to Implement Papers 12:46 - Making SinkSub Game with AI using ChatDEV 19:37 - Code Llama Paper & Models Grounded in Mathematical Truth, False Refusals 31:16 - The Only Path to Controllable AGI Paper Discussion (Max Tegmark) 40:55 - Baidu's ERNIE China's ChatGPT: Our Review 48:13 - Mike's Claude Prediction Comes True: Anthropic Release Claude PRO 50:35 - OpenAI Announce AI Developer Conference November 6th 2023 52:59 - Apple Siri AJAX Rumors: Apple ChatGPT? 54:25 - Time's 100 Most Influential People in AI LOLs SOURCES: ====== https://www.chinatalk.media/p/how-ernie-chinas-chatgpt-cracks-under https://openai.com/blog/announcing-openai-devday https://twitter.com/anthropicai/status/1699776481009053806?s=46&t=uXHUN4Glah4CaV-g2czc6Q https://www.theverge.com/2023/9/6/23861763/apple-ai-language-models-ajax-gpt-training-spending https://twitter.com/ylecun/status/1699779376961712473?s=46&t=uXHUN4Glah4CaV-g2czc6Q PAPERS: ====== https://arxiv.org/pdf/2307.07924v3.pdf https://arxiv.org/abs/2308.12950 https://arxiv.org/abs/2309.01933