Apple’s MM1 The New Recipe to Master AI Performance.
Plus, Cerebras WSE-3, the game-changing AI Chip enabling massive AI models, and Apple acquires Canadian AI startup DarwinAI
Hello Engineering Leaders and AI Enthusiasts!
Welcome to the 232nd edition of The AI Edge newsletter. This edition brings you “Apple’s MM1: The new recipe to master AI performance”
And a huge shoutout to our incredible readers. We appreciate you😊
In today’s edition:
🥘 Apple’s MM1: The new recipe to master AI performance
⚡ Cerebras WSE-3: AI chip enabling 10x larger models than GPT-4
🤖 Apple acquires Canadian AI startup DarwinAI
💡 Knowledge Nugget: ROI on GenAI might not be so great, after all by
Let’s go!
Apple’s MM1: The new recipe to master AI performance
Apple's MM1 AI model shows state-of-the-art language and vision capabilities. It was trained on a filtered dataset of 500 million text-image pairs from the web, including 10% text-only docs to improve language understanding.
The team experimented with different configurations during training. They discovered that using an external pre-trained high-resolution image encoder improved visual recognition. Combining different image, text, and caption data ratios led to the best performance. Synthetic caption data also enhanced few-shot learning abilities.
This experiment cements that using a blend of image caption, interleaved image text, and text-only data is crucial for achieving state-of-the-art (SOTA) few-shot results across multiple benchmarks.
Why does it matter?
Apple’s new model is promising, especially in developing image recognition systems for new categories or domains. This will help businesses and startups improve the speed of AI tool development specifically for text-to-image, document analysis, and enhanced visual recognition.
Cerebras WSE-3: AI chip enabling 10x larger models than GPT-4
Cerebras Systems has made a groundbreaking announcement unveiling its latest wafer-scale AI chip, the WSE-3. This chip boasts an incredible 4 trillion transistors, making it one of the most powerful AI chips on the market. The third-generation wafer-scale AI mega chip is twice as powerful as its predecessor while being power efficient.
The chip's transistor density has increased by over 50 percent thanks to the latest manufacturing technology. One of the most remarkable features of the WSE-3 chip is its ability to enable AI models that are ten times larger than the highly acclaimed GPT-4 and Gemini models.
Why does it matter?
The WSE-3 chip opens up new possibilities for tackling complex problems and pushing the boundaries of AI capabilities. This powerful system can train massive language models, such as the Llama 70B, in just one day. It will help enterprises create custom LLMs, rapidly reducing the time-to-market.
Apple acquires Canadian AI startup DarwinAI
Apple made a significant acquisition earlier this year by purchasing Canadian AI startup DarwinAI. Integrating DarwinAI's expertise and technology bolsters Apple's AI initiatives.
With this acquisition, Apple aims to tap into DarwinAI's advancements in AI technology, particularly in visual inspection during manufacturing and making AI systems smaller and faster. Leveraging DarwinAI's technology, Apple aims to run AI on devices rather than relying solely on cloud-based solutions.
Why does it matter?
Apple's acquisition of DarwinAI is a strategic move to revolutionize features and enhance its AI capabilities across various products and services. Especially with the iOS 18 release around the corner, this acquisition will help create new features and enhance the user experience.
Enjoying the daily updates?
Refer your pals to subscribe to our daily 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: ROI on GenAI might not be so great, after all
In a blog post,
discusses the potential limitations and challenges of achieving a high return on investment (ROI) with generative artificial intelligence (GenAI). The article highlights that GenAI has shown promise in various applications, but several factors can hinder its ROI.The blog further highlights the challenges of ROI with generative AI, like,
Accurately measuring the ROI of GenAI investments is a significant challenge.
Many GenAI benefits, such as productivity improvements and enhanced customer experience, have indirect or non-financial impacts.
Another factor impacting GenAI's ROI is the data quality used for training the AI models.
GenAI systems' performance heavily relies on the quality and diversity of the training data.
Furthermore, the article raises concerns about the potential risks and ethical implications associated with GenAI. It emphasizes the importance of addressing these concerns to ensure that GenAI investments' ROI is not overshadowed by negative consequences.
Why does it matter?
While GenAI can potentially deliver significant benefits and improve productivity, accurately measuring and achieving a high ROI can be challenging. This could be a barrier to entry for many companies in the AI-based market, where we are witnessing unprecedented startups and enterprises investing heavily in developing new tools.
What Else Is Happening❗
🤖 Microsoft expands the availability of Copilot across life and work.
Microsoft is expanding Copilot, its AI assistant, with the introduction of the Copilot Pro subscription for individuals, the availability of Copilot for Microsoft 365 to small and medium-sized businesses, and removing seat minimums for commercial plans. Copilot aims to enhance creativity, productivity, and skills across work and personal life, providing users access to the latest AI models and improved image creation. (Link)
💻 Oracle adds groundbreaking Generative AI features to its software
Oracle has added advanced AI capabilities to its finance and supply chain software suite, aimed at improving decision-making and enhancing customer and employee experience. For instance, Oracle Fusion Cloud SCM includes features such as item description generation, supplier recommendations, and negotiation summaries. (Link)
💰 Databricks makes a strategic investment in Mistral AI
Databricks has invested in Mistral AI and integrated its AI models into its data intelligence platform, allowing users to customize and consume models in various ways. The integration includes Mistral's text-generation models, such as Mistral 7B and Mixtral 8x7B, which support multiple languages. This partnership aims to provide Databricks customers with advanced capabilities to leverage AI models and drive innovation in their data-driven applications. (Link)
📱 Qualcomm emerges as a mobile AI juggernaut
Qualcomm has solidified its leadership position in mobile artificial intelligence (AI). It has been developing AI hardware and software for over a decade. Their Snapdragon processors are equipped with specialized AI engines like Hexagon DSP, ensuring efficient AI and machine learning processing without needing to send data to the cloud. (Link)
👓 MIT researchers develop peripheral vision capabilities for AI models
AI researchers are developing techniques to simulate peripheral vision and improve object detection in the periphery. They created a new dataset to train computer vision models, which led to better object detection outside the direct line of sight, though still behind human capabilities. A modified texture tiling approach accurately representing information loss in peripheral vision significantly enhanced object detection and recognition abilities. (Link)
New to the newsletter?
The AI Edge keeps engineering leaders & AI enthusiasts like you on the cutting edge of AI. From ML to ChatGPT to generative AI and LLMs, We break down the latest AI developments and how you can apply them in your work.
Thanks for reading, and see you tomorrow. 😊