- The Renaissance Times
- Posts
- Apple Enters the AI Arena
Apple Enters the AI Arena
Dear Artisan,
In the past year or so of AI buzz, there have been a few new names notably missing in terms of big tech players jumping into the AI game. Chief among those has been Apple.
That is all about to change.
Last month, Apple released their first foundation model, MM1, and have just followed it up with their latest model, ReALM (Reference Resolution As Language Modeling), which they claim beats OpenAI’s GPT-4.
Let’s dive in.
Hot Off The Press
Apple’s ReALM Challenges OpenAI’s GPT-4
DALL-E now lets you edit images in ChatGPT
India Creates its own AI safety coalition to deal with election misinformation
SiMa.ai secures $70M funding to introduce a multimodal GenAI chip
S&P Global launches AI benchmark for financial industry
The One Big Thing
One of the biggest questions surrounding the main players in AI has been, “where are Amazon and Apple?” Surely they have had some sort of play in the space they are waiting to make.
Apple seems to be playing its hand this year. Last month they announced their first foundation model, MM1, and the news has been rolling ever since. There are rumors of an acquisition of Peprplexity, the AI-based search engine, and talks of a partnership with Google to integrate their new foundation models into upcoming iPhones.
The latest announcement, however, was much more under the radar but could show just how far Apple is willing to go down this road. A few days ago Apple released the paper for ReALM (Reference Resolution As Language Modeling), which they claim beats GPT-4.
Apple says its latest AI model ReALM is even “better than OpenAI’s GPT4”.
It likely is as GPT4 has regressed because of “alignment”.
The ReALM war begins at WWDC 2024.
Paper: arxiv.org/pdf/2403.20329…
— Brian Roemmele (@BrianRoemmele)
8:50 PM • Apr 1, 2024
Now, this is just a paper so far and we are taking these results at Apple’s word. But the fact remains that this is the second model they have released within the span of a few weeks, and these models will likely live natively in various Apple devices in the future.
Most people will not need to have a massive LLM like GPT-4 on their iPhone, so it makes sense for Apple to experiment on and build out smaller, custom-fit models that serve specific purposes and match the exact devices they are made for.
Apple has a history of making the best tools for creators, and AI will be just that in the context of their products. Smaller, faster models that are hyper-optimized for the chips they run on, delivering an ideal experience.
This becomes especially important for devices like the Vision Pro, where hardware space is limited. This marks a huge step for Apple and indicates their seriousness in the space. Exciting times.
The Gallery
Finger lickin’ good!
Tools
Must have tools for every artisan to add to their toolkit:
Nomi: AI Personality with group chats
𝐃𝐞𝐬𝐢𝐠𝐧𝐄𝐝𝐢𝐭: A unified framework for precise and flexible image editing
Wanderboat AI: New travel planning tool
Meshy 2.5: A platform to create 3D content, texturing and modeling
Inkflow: A tool to generate structured non-fiction book manuscripts
Deep Tech
The newest and coolest in the research world that you need to know about:
Google presents Mixture-of-Depths: Dynamically allocating compute in transformer-based language models
Apple ReALM Technical Paper
Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction
Freditor: High-Fidelity and Transferable NeRF Editing by Frequency Decomposition
Language Models as Compilers: Simulating Pseudocode Execution Improves Algorithmic Reasoning in Language Models
Closing Thought
As the biggest tech companies go all in on AI, the decentralize AI movement will gain tons of steam
Work With Us!
The AI Renaissance is coming and we are building the best community of the people making it happen.
Contact us to sponsor your product or brand and reach the exact audience for your needs across our newsletter and podcast network.