AI Apocalypto Newsletter | Week 21
AI News This Week | AI Headlines!
Welcome to Week 21 of AI Apocalypto! Dive into the latest developments in AI ethics, emerging technologies, and controversial uses shaping our future. Stay informed with this week's headlines and insights from the AI frontier.
Ethical Usage of AI
Getty and Picsart introduce a licensed AI image generator, ensuring commercial safety and creator compensation.
California's Latest AI Safety Bill
• Discusses liability of AI developers for catastrophic AI risks.
• Supported by AI researchers leaders like Hinton and Bengio.
• Mandates safety testing for large-scale AI models.
• Critics argue it could stifle innovation.
• Proponents see it as necessary for public safety.
MIT's Machine Learning Method for Climate Modeling
• Generates high-resolution data efficiently.
• Integrates simplified physics and historical statistics.
• Aims to predict extreme weather impacts swiftly.
• Crucial for decision-making in vulnerable regions like Bangladesh.
Getty and Picsart Launch AI Image Generator
• AI generator trained on licensed stock images.
• Aims for commercial safety and full rights.
• Getty compensates creators annually.
• Model debuts later this year for Picsart subscribers.
Social Experiment with AI
• Used AI to generate apologies after insults.
• Subjected co-workers, family, friends to deceitful experiment.
• Aimed to explore AI's impact on sensitive human interactions.
• Apologized for tricking.
Unethical Usage of AI
Senthil Nayagam's deepfake campaign in India highlights the potential for electoral chaos fueled by AI advancements.
PDoom 70%: OpenAI's Risks Concerns
• Former OpenAI researcher Daniel Kokotajlo warns of a 70% chance of catastrophic AI harm.
• OpenAI accused of prioritizing AGI development over safety.
• OpenAI asserts commitment to safe AI and encourages debate.
Senthil Nayagam's Deepfake Campaign
• Created AI-generated videos using voice recording and pictures.
• Spread quickly, promoting or discrediting candidates.
• Despite ethical concerns, influenced India's recent elections.
• Highlights potential global electoral chaos from AI advancements.
Elon Musk's AI and Robotics Vision
• Nvidia staff warns of conflict between Musk's claims and actual orders.
• Musk aims to lead Tesla in AI and robotics.
• Promised 85,000 Nvidia H100 chips by year-end and $10 billion AI spending.
• Nvidia emails suggest Musk exaggerated procurement.
• Musk redirected GPU shipment to X, causing delays.
• Internal conflict and shareholder concerns.
Harvard AI Students' Concerns
• More worried about AI's extinction risks.
• Shorter "AGI timelines."
• Half believe AI will be as capable as humans within 30 years.
Meta's AI Training Plans Paused
• Plans to train AI systems using EU and UK user data halted.
• Objections from Irish Data Protection Commission and other authorities.
• Privacy policy change allowed AI training with user content.
Benjamin Miller's Job Changes
• Laid off, found new role at Undetectable AI.
• Replaced by AI-generated content in early 2023.
• Initially edited AI-produced outlines.
• Company develops software to make AI writing harder to detect.
AI Advancements
MIT's new machine learning method generates high-resolution climate data, crucial for predicting extreme weather impacts.
OpenAI Board Appoints Former NSA Head
• Concerns over OpenAI's board appointment.
• Paul M. Nakasone's appointment raises security and ethics questions.
• Edward Snowden warns of OpenAI's potential dangers.
WIRED Politics Lab: AI Chatbots Running for Elections
• Two AI chatbots, VIC and AI Steve, are running for elected office.
• VIC is campaigning for mayor in Cheyenne, Wyoming.
• AI Steve is running for Parliament in the UK.
• Reporter Vittoria Elliot interviews the bots and their supporters.
• David Gilbert discusses Google and Microsoft's AI chatbots' refusal to confirm the 2020 US election winner.
Kuaishou's Kling AI Model Outperforms OpenAI's Sora
• Creates 1080p, 30fps videos from a single prompt.
• Utilizes Diffusion Transformer architecture and advanced 3D reconstruction.
• Demonstrates China's AI leadership.
• OpenAI's Sora may lag behind China's advancements.
McGovern Institute Symposium Highlights Digital Technologies' Impact on Mental Health
• Emphasizes AI's potential for diagnosis and treatment.
• Discusses rising mental health challenges.
• Discusses AI-driven suicide risk prediction.
• Introduces new treatments like ketamine therapy and smartphone-based depression interventions.
Microsoft Deploys GPT-4 Model in Government-Only Network
• First major LLM operating separately from internet.
• Announced at AI Expo for National Competitiveness.
• Static server answers questions and writes code.
GPT-4 Enhances Reasoning with NLEPs
• Generates Python programs for math, data analysis, and symbolic reasoning.
• Improves accuracy, transparency, and efficiency.
• Avoids data privacy issues.
• Benefits from local processing of sensitive information.
• Future research aims for broader applications.
Yahoo Launches Revamped Yahoo News App
• Incorporates Artifact's AI-driven news delivery.
• Aims to enhance user engagement with tailored content.
• Offers AI-generated article summaries.
• Ad-supported, available on iOS and Android.
LinkedIn Introduces AI-Powered Features
• Resume/cover letter writing tools
• Learning personalization
• Improved search functionality
• Integration with Microsoft's AI initiatives
• Balancing innovation with regulatory compliance and authenticity.
University of Tokyo's Musashi Development
• Musashi, a humanoid robot, trained for electric car driving.
• Uses cameras for vision and mechanical hands for controls.
• Plans to enhance autonomous driving capabilities.
ETH Zurich's Autonomous Navigation System for Quadrupled Robots
• Enhances speed and maneuverability in urban settings.
• Uses reinforcement learning for smooth transitions.
• Tested in Zurich and Seville.
• Promising for robotic deliveries and urban logistics.