- Future Tools
- Posts
- Is Apple changing AI for good?
Is Apple changing AI for good?
Plus: AI video generation has a new player
Welcome back! Elon Musk is stirring the pot again—this time over the Apple and OpenAI partnership. In classic Musk fashion, he threatened to ban Apple devices entirely from company premises if the company integrates OpenAI at the operating system level, calling it an "unacceptable security violation."
Privacy concerns overblown? Maybe, but it certainly raises interesting questions about AI integration and security. What do you think?
Apple Unleashes a New Era of Intelligence
Wired
Apple’s Worldwide Developers Conference (WWDC) took over the AI world this week. Here's what you need to know.
Key announcements: At WWDC, Apple unveiled Apple Intelligence, a comprehensive suite of AI-powered features rolling out this fall to enhance the iPhone, Mac, and more. This update will include a more conversational Siri, custom AI-generated “Genmoji,” and integration with OpenAI’s ChatGPT for complex queries.
It’s all about making your devices smarter and more intuitive. Let’s dive deeper.
Siri x ChatGPT Integration
Siri is leveling up. When you ask Siri to complete a complex task (like creating a detailed trip itinerary), it’ll ask for your permission to bring in ChatGPT.
Key features:
Detailed information retrieval: Siri can now provide more comprehensive answers by leveraging ChatGPT’s extensive knowledge base. Instead of boilerplate answers, Siri will personalize responses.
Creative assistance: Siri can tap into ChatGPT’s creative capabilities, generating draft stories, poems, and even brainstorms.
Enhanced conversations: Siri's conversational abilities can maintain the context of previous queries, allowing for more natural and continuous conversations.
Images 🤝 AI
Apple Intelligence also introduced two image generation features:
Genmoji lets you create custom emoji-like reactions using AI.
Image Playground lets you generate images from text prompts in styles like animation and sketch.
Privacy-Focused AI
Apple Intelligence prioritizes privacy by running many features directly on your device. When additional processing power is needed, Apple’s Private Cloud Compute steps in without storing or accessing your data.
Siri now supports hundreds of in-app actions, such as updating contact information, retrieving images, and managing your calendar, all while keeping your data private.
Apple’s strategy? Create smaller, highly specialized models tailored to specific functions within its operating systems.
Smaller models like Apple Intelligence excel in speed and efficiency, especially for tasks handled directly on your device (for context: GPT-4 has over 1 trillion parameters compared to Apple’s ~3 billion).
They don’t cover as broad a range of functions as larger models, but they are optimized for the tasks that matter most to users (like face recognition and voice assistant).
Why it matters: Since the drop of ChatGPT in 2022, Apple has been eerily quiet on its AI strategy. Much speculation circled how Apple would integrate AI models into its tech. Now, the company is barreling out of the gates of WWDC with its answer: Apple Intelligence.
Apple Intelligence marks a significant advancement for AI, making sophisticated features more accessible and easily integrated into daily use.
Stability AI Releases Major Model Upgrade
Stability AI
Stability AI just released its most advanced text-to-image open model yet.
Some background: Stable Diffusion 3 Medium is the latest iteration in the Stable Diffusion series, an advanced set of text-to-image open models. Known for its robust text-to-image capabilities, this new model is designed to be more accessible, running efficiently on consumer-grade GPUs.
What makes SD3 Medium stand out?
Smaller but mighty: Unlike its 8 billion parameter counterpart SD3 Large, SD3 Medium packs a punch with just 2 billion parameters. This smaller size makes it highly efficient without sacrificing quality.
Performance on consumer hardware: SD3 Medium can run on 5GB of GPU VRAM, making it suitable for most consumer PCs and high-end laptops.
High-quality output: Despite its smaller size, SD3 Medium excels in producing photorealistic images, adhering closely to prompts, and handling typography and fine-tuning with ease. It delivers greater detail per megapixel thanks to its 16-channel Variational Autoencoder (VAE).
Want to try it for yourself? SD3 Medium is available for use through an API and the Stable Artisan service on Discord. Users can also download the model weights for non-commercial use from Hugging Face, allowing for further customization and experimentation.
Why it matters: The smaller size of SD3 Medium makes advanced text-to-image generation more accessible to everyday users. Because it’s capable of running on consumer hardware, it could set a new standard for text-to-image models for a broader audience.
Automate 50% of your workflow using AI tools (free masterclass) 🚀
Don’t pay for sub-par AI courses when you can learn it for FREE!
This incredible 3-hour Masterclass on AI & ChatGPT (worth $399) makes you a master of 25+ AI tools, hacks & prompting techniques to save 16 hours/week and do more with your time.
Get it now for absolutely free! (for first 100 users only) 🎁
This masterclass will teach you how to:
Do AI-driven data analysis to make quick business decisions
Make stunning PPTs & write content for emails, socials & more in minutes
Build AI assistants & custom bots in minutes
Solve complex problems, research 10x faster & make your simpler & easier
You’ll wish you knew about this FREE AI masterclass sooner 😉
The New Competitor in AI Video Generation
Kling
There's a new Sora competitor that you need to know about.
Kling is an emerging player in the AI video generation space, designed to rival well-known models like Sora. Developed by a team of Chinese AI developers, Kling is pushing the boundaries of video generation with enhanced features and quality.
What makes it different from Sora?
Advanced capabilities: Kling offers high-resolution video generation, improved rendering speeds, and a user-friendly interface. It’s designed to handle complex video generation tasks with greater efficiency and output quality than Sora.
One standout feature? Kling’s ability to generate videos with more precise control over lighting, textures, and transitions. This allows for more customized and professional-quality video content.
How to use it: Kling is available through a web-based platform. Upload your script, choose from a variety of templates, and let Kling do the rest. It also supports API integration for more advanced users.
Why it matters: Kling’s entry into the market (with advanced features and competitive pricing) could pressure US-based AI video model providers like OpenAI, Runway, and Pika to step up their game.
OpenAI will use Oracle’s chips to power more AI compute.
Luma AI debuts “Dream Machine” tech for super realistic video generation.
Apple passes Microsoft to become world’s most valuable company…again.
Google introduces Gemini Nano AI for the Pixel 8 Series in its latest feature drop.
Linq, which uses AI to automate a slew of tasks required for financial analysis and research, raises $6.6 million.
This is wild: A virtual rat with an AI brain mimics real rodent movement.
Apple is all-in on AI. Watch as I break down key updates from WWDC. Let’s just say…today’s top story is only the beginning.
Are closed-source or open-source AI models the future of artificial intelligence? On the latest episode of The Next Wave, we sit down with Matthew Berman to explore the pros and cons of closed models from giants like OpenAI and Google vs. open models like Llama and Mixtral.
And one more thing…your guide to this week’s huge wave of AI news, from yours truly.
That rounds out this week! Do you think Elon Musk's concerns are justified or unnecessary fanfare? Hit reply and let me know your thoughts!
—Matt (FutureTools.io)
P.S. This newsletter is 100% written by a human. Okay, maybe 96%.