🗞️ AI News Recap

Hey TEDchats Fam!

Happy Monday! In today’s AI News Recap, we bring you the latest developments and breakthroughs in the AI world, from groundbreaking new models and innovative tools to industry-shaking acquisitions and provocative statements. Stay informed and ahead of the curve with the most important AI news from the past week.

🐇 QUICK BITES — THE PAST WEEK

Anthropic Releases Its Best Model Yet: Claude 3.5 Sonnet

Available on the web and iOS app, Claude 3.5 Sonnet promises to surpass leading AI models like OpenAI’s GPT-4o and Google’s Gemini 1.5 in speed, accuracy, and versatility. Claude 3.5 Sonnet outperforms its predecessor and top rivals in multiple benchmarks, including seven out of nine general benchmarks and four out of five vision benchmarks. This chatbot offers enhanced capabilities in coding, workflow management, data interpretation, and humor understanding. The introduction of the Artifacts feature further elevates its functionality, allowing direct modification of Claude's responses. This feature represents a part of Anthropic AI's broader vision to transform conversational AI into a tool that supports collaborative work environments (see Anthropic’s announcement and key benchmarks here).

OpenAI CTO: 'Some Creative Jobs Maybe Will Go Away, But Maybe They Shouldn't Have Been There' — Public Uproar Ensues

During a recent lecture at Dartmouth’s School of Engineering, OpenAI CTO Mira Murati ignited a firestorm on the internet with her provocative statement: "Some creative jobs maybe will go away, but maybe they shouldn't have been there in the first place if the content that comes out of it is not very high quality." The public reaction was swift and fierce, sparking a social media uproar. The video’s comment section exploded with outraged responses such as, "What a revolting thing to say," "What a kick in the teeth," "Generative AI wouldn't exist without creatives' work," and "AI models would be nothing without those creative jobs." As of now, Murati has yet to respond to the backlash (see Dartmouth’s School of Engineering interview here; the controversial statement is at the 29-minute mark).

Runway Unveils Gen-3 Alpha for Improved Video Generation

Runway has launched Gen-3 Alpha, the first in an upcoming series of AI models that generate high-quality video clips from text descriptions and still images. This new model offers improved speed, fidelity, and control over video structure, style, and motion. While Gen-3 Alpha excels in creating expressive human characters, it currently supports clips up to 10 seconds long and faces challenges with complex interactions (see Runway’s announcement and demos here).

Wired Calls Out Perplexity as a 'Bullshit Machine' — Updated

Wired has conducted an investigation that heavily criticizes Perplexity, a free AI-powered answer engine that has previously gained attention. Now in the spotlight for negative reasons, Wired's investigation found that the AI search engine has been ignoring their efforts to prevent content scraping. Despite adding directives to their robots.txt files, a common method to instruct web crawlers to avoid certain parts of a site, Perplexity still accessed their content. Since the publication of Wired’s critique, in an ironic twist, Perplexity apparently plagiarized the critique itself! When questioned about these accusations, Perplexity CEO Aravind deflected the blame onto third-party content crawlers (read Wired's in-depth critique here and updated critique here).

TikTok Unveils AI-Generated Avatars for Enhanced Content Reach

TikTok is introducing new generative AI tools to help brands and content creators expand their global reach. Building on the Symphony ad suite, these tools include customizable digital avatars and advanced language dubbing features to break down language barriers and add a human touch to marketing content. The Symphony Digital Avatars come in two types: stock avatars based on diverse actors and custom avatars tailored to specific creators or brand representatives, capable of speaking multiple languages through AI dubbing. Videos will be marked as “AI-generated” to maintain transparency. Despite initial concerns about accuracy, this technology offers promising potential for enhancing global communication and setting new standards in digital marketing (see TikTok’s announcement and demo here).

Google DeepMind’s New AI Soundtrack Tool

Google DeepMind unveils a groundbreaking AI tool that generates soundtracks for videos using both video pixels and text prompts. This innovative technology can create drama scores, realistic sound effects, and synchronized dialogue, making it easier than ever to enhance video content with perfectly matched audio. Users can input specific text prompts or rely on the tool's ability to analyze video content to generate appropriate soundtracks automatically. With the capability to produce an unlimited number of audio variations, this tool promises to revolutionize how we pair audio with visual media (see Google DeepMind’s announcement).

Amazon Expands AI-Powered Listings to Europe

Amazon's generative AI tools for product listings are now available to sellers in Germany, Italy, France, Spain, and the U.K. These tools generate product descriptions, titles, and other details, and can enrich existing listings by adding missing information. While the rollout promises efficiency, Amazon advises sellers to review AI-generated content for accuracy due to potential hallucinations by large language models. This expansion comes at an interesting time, as just last week Meta paused plans to expand its AI training program to European markets due to the EU’s General Data Protection Regulation (see Amazon’s announcement).

YouTube Responds to AI Privacy Complaints

YouTube now allows users to request the removal of AI-generated content mimicking their face or voice. This update supports YouTube's commitment to responsible AI use. Users can report such content via the privacy request process, which YouTube will evaluate based on realism and identifiability (get more details in the YouTube Help Center here).

AI Visionary Sutskever Launches New Venture

Ilya Sutskever, former chief scientist and co-founder of OpenAI, has unveiled his latest project, Safe Superintelligence Inc. (SSI). In partnership with Daniel Gross and Daniel Levy, Sutskever aims to advance AI capabilities rapidly while ensuring safety remains paramount. Based in Palo Alto and Tel Aviv, SSI focuses solely on solving AI safety and superintelligence challenges. With a mission-driven approach, SSI plans to align safety and progress “through revolutionary engineering and scientific breakthroughs”. The company is currently recruiting technical talent to join their ambitious endeavor (see SSI’s announcement).

Apple Expands AI Education Through Its Developer Academy

Apple is revolutionizing its Developer Academy curriculum with a robust focus on AI training. Starting this fall, students and alumni will learn to build, train, and deploy machine learning models across Apple devices. This initiative spans 18 academies worldwide, including locations in Italy, Brazil, South Korea, Indonesia, Saudi Arabia, and the United States. With enhanced tools like Swift Assist and Swift 6, Apple aims to empower the next generation of developers to create innovative applications and drive technological advancements in their communities and beyond (see Apple’s announcement).

Snapchat Releases New AI Lenses as Live Image Filters

Snapchat’s upcoming on-device AI Lenses can transform your background and clothing in real time. By using filters that overlay digital effects on real-world images, it turns a simple text prompt like '50s sci-fi film' into an immersive augmented reality (AR) experience. With Snap providing an early preview, we should expect to see these innovative lenses in the coming months, along with more advanced tools for creators by year’s end. This exciting development also includes new AI tools in Lens Studio, allowing creators to generate 3D assets, custom face masks, and realistic transformations over users’ faces, bodies, and surroundings (see Snap’s preview).

Luma AI's Latest Video Tool: Impressive Innovation or Pixar Rip-Off?

Luma AI has launched an innovative tool called the Dream Machine, which allows users to generate high-quality video clips from simple text prompts in a matter of minutes. After its announcement last week, Luma AI shared several videos produced using the Dream Machine, including an animated story named "Monster Camp". However, the release sparked controversy when viewers noticed a character resembling Mike Wazowski from Pixar’s Monsters, Inc., appearing at about 56 seconds into the video. Observers pointed out that multiple characters and the overall look seemed to draw heavily from the Pixar franchise, leading to a flurry of questions. So far, neither Luma AI nor Disney has commented publicly on the issue (see Luma AI’s video here).

OpenAI Acquires Data-Driven Startup Rockset in One of Its Largest Startup Purchases

OpenAI has acquired Rockset, a leading real-time analytics database, to enhance data indexing and querying capabilities. This integration aims to help users, developers, and enterprises better leverage their data and access real-time information through AI products. Rockset’s team will join OpenAI to support this effort. OpenAI COO Brad Lightcap and Rockset CEO Venkat Venkataramani expressed excitement about transforming data into actionable intelligence. While the exact numbers have not been disclosed, it appears OpenAI acquired Rockset in a stock deal valuing the company at several hundred million dollars, making it one of OpenAI's largest acquisitions. This acquisition aligns with OpenAI’s strategy to strengthen its enterprise sales and technology divisions (see OpenAI’s announcement).

🧰 AI TOOLBOX — RECAP

👋 UNTIL NEXT TIME

That's all for this week's AI News Recap. Stay tuned for more updates and insights as we continue to explore the ever-evolving world of artificial intelligence. Thank you for reading, and we'll see you on Wednesday with more exciting AI news!