Last Week in AI Skynet Today

Our 164th episode with a summary and discussion of last week's big AI news!
Read out our text newsletter and comment on the podcast at https://lastweekin.ai/
Email us your questions and feedback at contact@lastweekin.ai and/or hello@gladstone.ai
Timestamps + links:
Tools & Apps
(00:04:02) Meta, in Its Biggest A.I. Push, Places Smart Assistants Across Its Apps
(00:07:26) Microsoft launches Phi-3, its smallest AI model yet
(00:15:35) The Ray-Ban Meta Smart Glasses have multimodal AI now
(00:17:32) OpenAI winds down AI image generator that blew minds and forged friendships in 2022
(00:18:44) Baidu claims 200 million users for Ernie chatbot after only 13 months
(00:21:13) The new Adobe Photoshop gets an in-app image generator, major Generative Fill upgrades

Applications & Business
(00:22:22) Intel & The Pentagon Deepen Ties To Develop World’s Most Advanced Chips
(00:27:58) Meta Says It Plans to Spend Billions More on A.I.
(00:31:36) OpenAI CEO Sam Altman invests in solar power firm Exowatt to fuel AI datacenters
(00:33:58) Google consolidates AI-focused DeepMind, Research teams
(00:36:22) Microsoft and OpenAI bet $100 billion to free themselves from the shackles and overreliance on the world's most profitable semiconductor chip brand for AI chips

Projects & Open Source
(00:39:03) Apple releases OpenELM: small, open source AI models designed to run on-device
(00:44:12) Snowflake launches Arctic, an open ‘mixture-of-experts’ LLM to take on DBRX, Llama 3

Research & Advancements
(00:48:08) The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions
(00:55:11) Groq’s breakthrough AI chip achieves blistering 800 tokens per second on Meta’s LLaMA 3
(00:59:52) Microsoft shows off VASA-1, an AI framework that makes human headshots talk, sing
(01:01:59) Intel Builds World’s Largest Neuromorphic System to Enable More Sustainable AI

Policy & Safety
(01:05:11) Deepfakes of Bollywood stars spark worries of AI meddling in India election
(01:08:51) LLM Agents can Autonomously Exploit One-day Vulnerabilities
(01:15:27) The Necessity of AI Audit Standards Boards
(01:19:45) A Mechanism-Based Approach to Mitigating Harms from Persuasive Generative AI
(01:22:45) COERCING LLMS TO DO AND REVEAL (ALMOST) ANYTHING
(01:26:40) China acquired recently banned Nvidia chips in Super Micro, Dell servers, tenders show

Synthetic Media & Art
(01:29:08) Drake threatened with lawsuit over diss track featuring AI Tupac

Our 163rd episode with a summary and discussion of last week's big AI news!
Note: apology for this one coming out a few days late, got delayed in editing it -Andrey
Read out our text newsletter and comment on the podcast at https://lastweekin.ai/
Email us your questions and feedback at contact@lastweekin.ai and/or hello@gladstone.ai
Timestamps + links:
Intro / Banter
Tools & Apps
(00:02:16) Meta releases Llama 3, claims it’s among the best open models available
(00:14:01) Elon Musk’s xAI Unveils Grok-1.5 Vision, Beats OpenAI’s GPT-4V
(00:17:55) Reka releases Reka Core, its multimodal language model to rival GPT-4 and Claude 3 Opus
(00:21:50) Cohere Compass Private Beta: A New Multi-Aspect Embedding Model
(00:23:48) Amazon Music’s Maestro lets listeners make AI playlists
(00:24:36) Snap plans to add watermarks to images created with its AI-powered tools

Applications & Business
(00:25:52) Boston Dynamics unveils new Atlas robot for commercial use
(00:30:32) TSMC’s $65 billion bet still leaves US missing piece of chip puzzle
(00:36:30) U.S. blacklists Intel's and Nvidia's key partner in China — three other Chinese firms also included in the blacklist for helping the military
(00:38:37) Elon Musk says the next-generation Grok 3 model will require 100,000 Nvidia H100 GPUs to train
(00:40:22) Dr. Andrew Ng appointed to Amazon’s Board of Directors
(00:41:55) Collaborative Robotics Locks Up $100M, Latest Robot Startup To Raise Big

Projects & Open Source
(00:44:08) OpenEQA: Embodied Question Answering in the Era of Foundation Models
(00:50:03) Introducing Idefics2: A Powerful 8B Vision-Language Model for the community

Research & Advancements
(00:51:21) RHO-1: Not All Tokens Are What You Need
(00:57:21) Scaling Laws for Fine-Grained Mixture of Experts
(01:03:20) Chinchilla Scaling: A replication attempt
(01:07:18) China develops new light-based chiplet that could power artificial general intelligence — where AI is smarter than humans
(01:10:45) OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments

Policy & Safety
(01:13:44) U.S. Commerce Secretary Gina Raimondo Announces Expansion of U.S. AI Safety Institute Leadership Team
(01:17:18) NSA Publishes Guidance for Strengthening AI System Security
(01:19:19) Foundational Challenges in Assuring Alignment and Safety of Large Language Models
(01:24:11) Former OpenAI Board Member Calls for Audits of Top AI Companies
(01:27:35) SoA survey reveals a third of translators and quarter of illustrators losing work to AI

Synthetic Media & Art
(01:30:25) Medium bans AI-generated content from its paid Partner Program

Our 162nd episode with a summary and discussion of last week's big AI news!
Read out our text newsletter and comment on the podcast at https://lastweekin.ai/
Email us your questions and feedback at contact@lastweekin.ai and/or hello@gladstone.ai
Timestamps + links:
Tools & Apps
(00:02:50) AI-Music Arms Race: Meet Udio, the Other ChatGPT for Music
(00:07:42) Anthropic launches external tool use for Claude AI, enabling stock ticker integrations and more
(00:11:51) Building LLMs for Code Repair
(00:14:16) Early Reviews of Humane AI Pin Aren’t Impressed
(00:16:23) Microsoft 365’s Copilot gets a GPT-4 Turbo upgrade and improved image generation
(00:18:41) AI editing tools are coming to all Google Photos users

Applications & Business
(00:19:21) Google announces the Cloud TPU v5p, its most powerful AI accelerator yet
(00:23:32) Meta unveils its newest custom AI chip as it races to catch up
(00:27:27) Intel Unveils New AI Accelerator in Bid to Challenge Nvidia
(00:30:46) Adobe Is Buying Videos for $3 Per Minute to Build AI Model
(00:32:55) OpenAI transcribed over a million hours of YouTube videos to train GPT-4
(00:36:23) Waymo will launch paid robotaxi service in Los Angeles on Wednesday
(00:37:23) OpenAI removes Sam Altman's ownership of its Startup Fund

Projects & Open Source
(00:39:51) Mistral AI Stuns With Surprise Launch of New Mixtral 8x22B Model
(00:43:54) Google updates its Gemma AI model family with variants for coding and research
(00:47:04) Aurora-M: The First Open Source Multilingual Language Model Red-teamed according to the U.S. Executive Order

Research & Advancements
(00:52:08) Mixture-of-Depths: Dynamically allocating compute in transformer-based language models
(00:57:41) Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention
(01:03:31) Octopus v2: On-device language model for super agent
(01:07:54) Bigger is not Always Better: Scaling Properties of Latent Diffusion Models
(01:09:54) Many-shot Jailbreaking

Policy & Safety
(01:15:08) Schiff unveils AI training transparency measure
(01:20:25) Linwei Ding was a Google software engineer. He was also a prolific thief of trade secrets, say prosecutors.
(01:26:11) Responsible Reporting for Frontier AI Development
(01:30:08) US govt wants to talk to tech companies about AI electricity demands — eyes nuclear fusion and fission
(01:32:39) Washington state judge blocks use of AI-enhanced video as evidence in possible first-of-its-kind ruling
(01:36:45) Trudeau announces $2.4 billion for AI-related investments

Synthetic Media & Art
(01:39:26) Billie Eilish, Pearl Jam, Nicki Minaj Among 200 Artists Calling for Responsible AI Music Practices

Fun!
(01:41:52) OpenAI's Sora just made its first music video and it's like a psychedelic trip

Our 161st episode with a summary and discussion of last week's big AI news!
Check 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’s a video version on YouTube.
Read out our text newsletter and comment on the podcast at https://lastweekin.ai/
Email us your questions and feedback at contact@lastweekin.ai and/or hello@gladstone.ai
Note - one extra story we didn't get to but worth knowing from this week: ‘Totally surreal’: OpenAI shares first short films created with new AI tool Sora
Timestamps + links:
(00:00:00) Intro / Banter
Tools & Apps
(00:05:20) Google starts testing AI overviews from SGE in main Google search interface
(00:10:00) Adobe’s new GenStudio platform is an AI factory for advertisers
(00:13:53) Claude-3 Haiku has reached GPT-4 level by our user preference
(00:15:26) Microsoft Teams is getting smarter Copilot AI features
(00:17:16) Samsung is beating Apple in the race to bring AI to smartphones
(00:19:18) Elon Musk says all premium subscribers on X will gain access to AI chatbot Grok this week

Applications & Business
(00:22:36) Stability AI CEO resigns to ‘pursue decentralized AI’
(00:26:43) Amazon spends $2.75 billion on AI startup Anthropic in its largest venture investment yet
(00:31:16) Intel Gaudi 2 Accelerators Showcase Competitive Performance Per Dollar Against NVIDIA H100 In MLPerf 4.0 GenAI Benchmarks
(00:33:43) Chip Startup Celestial AI Lands Massive $175M Series C
(00:36:10) Accenture Invests in Sanctuary AI to Bring AI-Powered, Humanoid Robotics to Work Alongside Humans
(00:37:44) Humanoid robots are joining the Mercedes-Benz workforce

Projects & Open Source
(00:38:40) Introducing DBRX: A New State-of-the-Art Open LLM
(00:44:06) Common Corpus: A Large Public Domain Dataset for Training LLMs
(00:46:07) DROID: A Large-Scale In-the-Wild Robot Manipulation Dataset
(00:48:15) InternLM2 Technical Report

Research & Advancements
(00:51:49) TacticAI: an AI assistant for football tactics
(00:55:44) On the Conversational Persuasiveness of Large Language Models: A Randomized Controlled Trial
(00:59:26) AutoDev: Automated AI-Driven Development
(01:02:48) Reverse Training to Nurse the Reversal Curse
(01:07:43) Cobra: Extending Mamba to Multi-Modal Large Language Model for Efficient Inference
(01:09:48) Gamba: Marry Gaussian Splatting with Mamba for single view 3D reconstruction

Policy & Safety
(01:11:12) General Assembly adopts landmark resolution on artificial intelligence
(01:16:24) Israel Deploys Expansive Facial Recognition Program in Gaza
(01:19:49) LINEARITY OF RELATION DECODING IN TRANSFORMER LANGUAGE MODELS
(01:24:52) US Weighs Sanctioning Huawei’s Secretive Chinese Chip Network
(01:27:51) New York City welcomes robotaxis — but only with safety drivers
(01:28:40) The White House Puts New Guardrails on Government Use of AI

Synthetic Media & Art
(01:31:17) BBC Will Stop Using AI For ‘Doctor Who’ Promotion After Receiving Complaints

Our 160th episode with a summary and discussion of last week's big AI news!
Read out our text newsletter and comment on the podcast at https://lastweekin.ai/
Email us your questions and feedback at contact@lastweekin.ai and/or hello@gladstone.ai
Timestamps + links:
Tools & Apps
(00:01:36) Adobe Substance 3D’s AI features can turn text into backgrounds and textures
(00:05:04) OpenAI’s chatbot store is filling up with spam
(00:11:02) Apple’s AI ambitions could include Google or OpenAI

Applications & Business
(00:13:31) Nvidia reveals Blackwell B200 GPU, the ‘world’s most powerful chip’ for AI
(00:19:34) Microsoft to Pay Inflection AI $650 Million After Scooping Up Most of Staff
(00:24:33) Figure 01: Conversations & Actions in Humanoid Robotics!
(00:28:07) OpenAI's GPT-4.5 Turbo leaked on search engines and could launch in June
(00:30:32) Abu Dhabi in talks to invest in OpenAI chip venture
(00:33:43) Nvidia Announces GR00T, a Foundation Model For Humanoids

Projects & Open Source
(00:35:38) Open Release of Grok-1
(00:41:25) Stability AI brings a new dimension to video with Stable Video 3D
(00:44:23) Colossal-AI Team Introduces Open-Sora: An Open-Source Library for Video Generation
(00:45:43) Evolutionary Optimization of Model Merging Recipes

Research & Advancements
(00:46:52) DiPaCo: Distributed Path Composition
(00:53:58) MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training
(00:59:37) PERL: Parameter Efficient Reinforcement Learning from Human Feedback
(01:01:55) VideoAgent: Long-form Video Understanding with Large Language Model as Agent
(01:05:38) MusicHiFi: Fast High-Fidelity Stereo Vocoding

Policy & Safety
(01:06:44) The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearning
(01:12:43) Exclusive: U.S. Must Move ‘Decisively’ to Avert ‘Extinction-Level’ Threat From AI, Government-Commissioned Report Says
(01:18:18) Evaluating Frontier Models for Dangerous Capabilities
(01:22:55) Chinese and western scientists identify ‘red lines’ on AI risks
(01:26:11) Google fined $272M by French government over AI use of news content
(01:27:58) Elvis Act Signed Into Tennessee Law to Protect Musicians From AI Deepfakes

Synthetic Media & Art
(01:30:17) AI-Generated Science
(01:35:35) YouTube adds new AI-generated content labeling tool

Fun!
(01:37:29) 10 of My Most Popular Text-To-Image Series (+Prompts)

Our 159th episode with a summary and discussion of last week's big AI news!
Check 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’s a video version on YouTube.
Read out our text newsletter and comment on the podcast at https://lastweekin.ai/
Email us your questions and feedback at contact@lastweekin.ai and/or hello@gladstone.ai
Timestamps + links:
Intro / Banter
Tools & Apps
(00:03:24) Inflection-2.5: meet the world's best personal AI
(00:06:37) Introducing Devin, the first AI software engineer
(00:11:00) DoorDash’s new AI-powered ‘SafeChat+’ tool automatically detects verbal abuse
(00:12:44) Anthropic releases Claude 3 Haiku, an AI model built for speed and affordability
(00:13:30) Pika Labs just added sound effects to its generative AI videos — here’s how it sounds
(00:15:33) Salesforce announces new AI tools for doctors

Applications & Business
(00:17:33) Sam Altman Rejoins OpenAI Board Along With Three New Directors
(00:21:15) Cohere releases powerful ‘Command-R’ language model for enterprise use
(00:23:16) Building Meta’s GenAI Infrastructure
(00:25:53) Baidu Launches China's First 24/7 Robotaxi Service

Projects & Open Source
(00:26:54) Croissant: a metadata format for ML-ready datasets
(00:29:40) SaulLM-7B: A pioneering Large Language Model for Law
(00:31:45) Kai-Fu Lee’s AI Company “01.AI” Announces the Open Source of the Yi-9B Model

Research & Advancements
(00:33:50) A generalist AI agent for 3D virtual environments
(00:39:16) Stealing Part of a Production Language Model
(00:42:01) Data Interpreter: An LLM Agent For Data Science
(00:43:54) ShortGPT: Layers in Large Language Models are More Redundant Than You Expect
(00:44:55) PixArt-Σ: Weak-to-Strong Training of Diffusion Transformer for 4K Text-to-Image Generation

Policy & Safety
(00:46:24) World’s first major act to regulate AI passed by European lawmakers
(00:48:57) US spearheads first UN resolution on artificial intelligence — aimed at ensuring equal access
(00:51:27) Google restricts election-related queries for its Gemini chatbot

Synthetic Media & Art
(00:52:43) Researchers tested leading AI models for copyright infringement using popular books, and GPT-4 performed worst
(00:55:27) Nvidia Says NeMo AI Platform Complies With Copyright After Authors’ Complaint
(00:57:23) Five of this year’s Pulitzer finalists are AI-powered

Fun!
(00:58:12) I made by Superman action figure talk with Pika Labs’ new AI lip sync tool — watch this