AI Glasses That See Through Your Privacy
Plus: Meta launches Movie Gen, OpenAI has introduced canvas, OpenAI just secured $6.6B funding, and more.
Hello Engineering Leaders and AI Enthusiasts!
This newsletter brings you the latest AI updates in just 4 minutes! Dive in for a quick summary of everything important that happened in AI over the last week.
And a huge shoutout to our amazing readers. We appreciate you😊
In today’s edition:
👓 AI glasses that see through your privacy
🎬 Meta launches Movie Gen for advanced media generation
🎨 OpenAI's canvas is changing how we collab with AI
💰 OpenAI just secured $6.6B funding
🤖 OpenAI's four major announcements on DevDay
🚫 Newsom hits pause on California's AI safety bill
📚 Knowledge Nugget: AI Creativity Is A Question of Quality, Not Novelty by
Let’s go!
AI glasses that see through your privacy
Harvard students presented a system using Meta's Ray-Ban smart glasses that reveals personal information about strangers. This proof-of-concept has sparked significant privacy concerns. These smart glasses can reveal your name, address, and phone number just by looking at you. I-XRAY, built by AnhPhu Nguyen and Caine Ardayfio, combines face recognition, large language models, and public databases.
But don't panic just yet. The creators aren't releasing this digital X-ray vision to the public. Instead, they're sounding the alarm about misusing current tech. They've even provided a handy guide on scrubbing your info from the data sources I-XRAY taps into, like Pimeyes and FastPeopleSearch.
Why does it matter?
I-XRAY shows how easily AI can stitch our online footprints into a comprehensive profile – without our consent. This project is a wake-up call to make stronger privacy laws and ethical AI development.
Meta launches Movie Gen for advanced media generation
Meta just dropped Movie Gen, a suite of advanced models directly competing with OpenAI's Sora. It comprises four distinct models: a large-scale 30B model for video generation, a 13B model dedicated to audio processing, a specialized model for personalized video creation, and a separate model designed for video editing tasks.
The system can create high-definition videos from text, edit existing clips, personalize videos with your face, and even whip up custom soundtracks. It has video personalization and precise editing capabilities. We're talking 1080p HD videos up to 16 seconds long, complete with synced 48kHz audio—all from a simple text prompt.
Why does it matter?
Mark Zuckerberg's post shows that Movie Gen is everyone's mini-Hollywood studio. As these models are integrated into Instagram, they could transform content creation, providing users with a sophisticated video editing tool that only requires text prompts to operate.
OpenAI's canvas is changing how we collab with AI
OpenAI just launched "canvas," an interface for ChatGPT that will change how we collaborate with AI on writing and coding projects. This separate window allows users to work side-by-side with ChatGPT, offering inline feedback, targeted editing, and a suite of handy shortcuts for tasks like adjusting text length or debugging code.
(Source)
In testing, accuracy was boosted by 30% and quality by 16% compared to the standard ChatGPT experience. Currently in beta for ChatGPT Plus and Team users, OpenAI plans to roll it out to all users once it's ready for prime time.
Why does it matter?
This is more than just a cosmetic change. Making AI assistance more contextual, intuitive, and project-oriented could be the push needed to mainstream AI collaboration with more refined interactions.
OpenAI just secured $6.6B funding
OpenAI secured a $6.6 billion funding round, valuing it at $157 billion. Led by Thrive Capital and backed by Microsoft and Nvidia, the deal comes with a twist: OpenAI is nudging investors to avoid rival AI startups.
Despite projected $5 billion losses this year, executive departures, and a lawsuit from Elon Musk, OpenAI plans to use the funds to expand computing capacity and develop more advanced AI tools.
Why does it matter?
This is like watching a tech version of Game of Thrones, where OpenAI is amassing power to rule the AI kingdom. By limiting investor options, OpenAI will stifle innovation from smaller players.
OpenAI's four major announcements on DevDay
OpenAI launched four key updates on DevDay 2024. These innovations aim to make AI more accessible and affordable for developers and businesses.
Realtime API enables low-latency, multimodal experiences and supports natural speech-to-speech conversations with six preset voices.
Model Distillation lets developers automatically store completions, evaluate performance, and seamlessly integrate fine-tuning—all within the OpenAI platform.
Prompt Caching is designed to optimize costs and processing times for developers using AI APIs, particularly for recurring or repeated inputs.
GPT-4o's new vision fine-tuning update enables developers to train AI models using both images and text, enhancing image comprehension.
Why does it matter?
These updates significantly lower the barrier to entry for AI development. By prioritizing cost reduction and developer tools, OpenAI is paving the way for smaller companies and startups to use its powerful models.
Newsom hits pause on California's AI safety bill
California Governor Gavin Newsom vetoed a sweeping AI safety bill that would have required large AI models to undergo safety testing before deployment. The bill, supported by Elon Musk and leading researchers, faced fierce opposition from tech giants like OpenAI and Google, as well as prominent Democrats in Congress.
Newsom argued the bill's approach was too broad, applying "stringent standards to even the most basic functions." Instead, he committed to formulating new legislation with academic experts and expanding workplace AI applications through pilot projects with state agencies.
Why does it matter?
This veto stresses the ongoing struggle to balance AI innovation with safety concerns. As California often sets de facto national standards for tech regulation, Newsom's decision could significantly influence the trajectory of AI governance across the US.
Enjoying the latest AI updates?
Refer your pals to subscribe to our newsletter and get exclusive access to 400+ game-changing AI tools.
When you use the referral link above or the “Share” button on any post, you'll get the credit for any new subscribers. All you need to do is send the link via text or email or share it on social media with friends.
Knowledge Nugget: AI Creativity Is A Question of Quality, Not Novelty
In this thought-provoking article,
argues that the real challenge for AI creativity isn't generating new content but producing high-quality work that meets complex constraints. He points out that while AI can easily create novel outputs (like random numbers), true creativity involves satisfying multiple difficult constraints simultaneously—think balancing humor and tension in a movie like Ghostbusters.The piece suggests we should judge AI's creativity by its ability to meet domain-specific quality standards rather than debating whether it's "truly creative." Examples like chess AIs show that AI can already produce creative and valuable outputs that influence human strategies in narrow fields.
Why does it matter?
This perspective shifts the AI creativity debate from philosophical arguments to practical assessments of AI's abilities in different fields. It provides a framework for evaluating AI advancements, like OpenAI's o1 model, to accelerate progress in AI-assisted creative and scientific works.
What Else Is Happening❗
💼 Inflection AI launches enterprise system with Intel, offering cloud service, API, and future local appliance for businesses.
🤖 OpenAI's case study on Altera shows GPT-4-powered AI agents excel at natural interactions and shows superior performance in Minecraft-based tests.
💊 Cleveland Clinic and IBM develop AI model predicting drug-microbe-pain receptor interactions, advancing non-addictive pain treatments.
📢 Google introduced ads to its AI search summaries while launching new AI features, including video analysis and voice input capabilities in Google Lens.
🚀 Black Forest Labs launched Flux 1.1 Pro, an enhanced text-to-image AI model 6x faster than its predecessor and outperformed competitors like Midjourney and DALL-E.
👵 MIT researchers created "Future You," an AI system that lets users converse with and question a simulated version of their older selves.
🔬 Google is developing advanced reasoning AI models that can solve complex, multi-step problems to rival OpenAI's o1, but it is taking a cautious approach to releasing them.
🎯 OpenAI Head of Product highlights real-time API's potential for voice AI interactions, pricing at ~30¢/minute for actual speech.
🖥️ Microsoft announced AI upgrades to Copilot with new vision, voice, and personalization features, reintroducing the controversial Recall feature.
🚀 Liquid AI introduces Liquid Foundation Models (LFMs), rivaling transformers with high performance and efficiency in smaller models.
New to the newsletter?
The AI Edge keeps engineering leaders & AI enthusiasts like you on the cutting edge of AI. From machine learning to ChatGPT to generative AI and large language models, we break down the latest AI developments and how you can apply them in your work.
Thanks for reading, and see you next week! 😊