BROADCASTS.com

Our 151st episode with a summary and discussion of last week's big AI news!\nCheck out our sponsor, the SuperDataScience podcast. You can listen to SDS across all major podcasting platforms (e.g., Spotify, Apple Podcasts, Google Podcasts) plus there\u2019s a video version on YouTube.\nRead out our text newsletter and comment on the podcast at https://lastweekin.ai/\nEmail us your questions and feedback at contact@lastweekin.ai\nTimestamps + links:\n(00:00:00) Intro / Banter\nTools & Apps(00:02:55) Microsoft\u2019s new Copilot Pro brings AI-powered Office features to the rest of us\n(00:04:43) Microsoft Copilot is now using the previously-paywalled GPT-4 Turbo, saving you $20 a month\n(00:07:02) Amazon launches generative AI tool to answer shoppers\u2019 questions\n\nApplications & Business(00:08:07) DeepMind spin-off aims to halve drug discovery times following Big Pharma deals\n(00:13:16) China planning 1,600-core chips that use an entire wafer \u2014 similar to American company Cerebras 'wafer-scale' designs\n(00:17:12)\xa0 ChatGPT will have video functionality and more accuracy in future versions \u2014 Sam Altman says GPT-5 will be a big improvement\n(00:19:38) Huawei Teardown Shows 5nm Chip Made in Taiwan, Not China\n(00:22:15) OpenAI\u2019s news publisher deals reportedly top out at $5 million a year\n(00:25:04) Inside AI Unicorn Anthropic\u2019s Unusual $750 Million Fundraise\n(00:28:46) Microsoft executive Dee Templeton joins OpenAI board - Bloomberg News\n(00:31:33) OpenAI-backed 1X raises another $100M for the race to humanoid robots\n\nProjects & Open Source(00:34:25) Meet LLama.cpp: An Open-Source Machine Learning Library to Run the LLaMA Model Using 4-bit Integer Quantization on a MacBook\n(00:38:06) LLAMA PRO: Progressive LLaMA with Block Expansion\n(00:43:48) One of the world\u2019s largest AI training datasets is about to get bigger and \u2018substantially better\u2019\n\nResearch & Advancements(00:47:52) Towards Conversational Diagnostic AI\n(00:51:50)\xa0 Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training\n(00:56:27) LLM Augmented LLMs: Expanding Capabilities through Composition\n(01:02:23) Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models\n(01:05:16) Clinical predictive models created by AI are accurate but study-specific, researchers find\n(01:09:01) Secure, Governable Chips\n\nPolicy & Safety(01:14:01) US companies and Chinese experts engaged in secret diplomacy on AI safety\n(01:18:28) China Cyberspace Security Association releases first batch of Chinese basic corpus\n(01:22:33) OpenAI Bans Use of AI Tools for Campaigning, Voter Suppression\n(01:24:49) OpenAI Quietly Deletes Ban on Using ChatGPT for \u201cMilitary and Warfare\u201d\n\nSynthetic Media & Art(01:27:20) Musicians Set to Begin Contract Negotiations With Studios On AI, Streaming Priorities\n(01:29:26) Scammy AI-Generated Book Rewrites Are Flooding Amazon\n(01:33:15) Deepfaked Celebrity Ads Promoting Medicare Scams Run Rampant on YouTube\n\n(01:35:15) Outro