AI can predict Alzheimer's years earlier

Hello everyone! Here’s what you need to know about AI today:

👉 Scientists can now detect Alzheimer’s years before any symptoms appear

👉 Google released a new embedding AI model

👉 X added the feature to ask Grok by mentioning it in replies

and many more!

📧 Did someone forward you this email? Subscribe here for free to get the latest AI news everyday!

Read time: 5 minutes

What’s going on: Researchers at Mass General Brigham have developed a new AI tool that can predict brain/cognitive decline in patients years before symptoms appear, potentially revolutionizing dementia treatment. The AI analyzes subtle changes in brain activity during sleep using electroencephalography (EEG).

What does it mean: The study trained the AI using sleep study data from women over 65, tracked over five years. The AI tool correctly identified 85% of individuals who experienced cognitive decline, with an overall accuracy of 77%. This early detection system could provide valuable time for at-risk individuals to implement lifestyle changes like exercise, mental activity, and a balanced diet, potentially mitigating the effects of cognitive impairment.

More details: 

• The researchers studied 281 women (185 controls, 96 developed impairment) aged 82+ from the Sleep and Cognition Study cohort and analyzed 4-channel EEG recordings during REM/NREM sleep stages.

• Features from the N3 sleep stage and gamma band had the strongest predictive power. Adding demographics, sleep macrostructure, and APOE genotype did not enhance prediction performance.

• Interested in this study? Read their published paper in the Journal of Alzheimer's Disease.

What’s going on: Google has introduced a new experimental text embedding model, Gemini Embedding, as part of its Gemini developer API. This model, trained using Google's Gemini family of AI models, translates text into numerical representations that capture semantic meaning, making it useful for applications like document retrieval and classification. Google claims that Gemini Embedding surpasses the performance of its previous model, text-embedding-004, and demonstrates competitive performance on industry benchmarks.

What does it mean: The new model boasts several advantages, including the ability to handle larger text and code inputs and support for over 100 languages, doubling the language support of its predecessor. Google emphasizes Gemini Embedding's broad applicability across diverse fields like finance, science, legal, and search, attributing this to its enhanced understanding of language and nuanced context.

More details: 

• This model is now in experimental phase. Google plans to release a stable, generally available version of Gemini Embedding in the coming months.

• This release positions Google alongside companies like Amazon (Titan), Cohere, and OpenAI in offering embedding models through their APIs.

• Embeddings enable a wide range of applications such as efficient retrieval (large databases, enterprise search, and legal documents), Retrieval-Augmented Generation (RAG), clustering and categorization, classification, and text similarity.

• Want to see the benchmarks and learn about the details of Gemini Embedding? Read Google’s official blog post.

What’s going on: Pharmaceutical giants like Johnson & Johnson, Merck, and Eli Lilly are increasingly embracing AI to enhance efficiency in drug development. These companies are investing in AI training programs for their employees, aiming to equip them with the skills needed to leverage AI technologies effectively.

What does it mean: The adoption of generative AI is expected to significantly impact the pharmaceutical industry by accelerating the identification of new therapies, optimizing clinical trials, and refining marketing strategies. While traditional expertise in biology and chemistry remains important, newer roles like AI engineers may require a blend of traditional knowledge and AI skills. The implementation of generative AI has the potential to save the pharmaceutical industry billions of dollars annually through increased productivity.

More details:

• Johnson & Johnson is focusing on creating bilingual employees who are proficient in both their specialized job functions and AI.

• Merck has developed its own AI platform, GPTeal, to enable employees to use AI tools for various tasks while ensuring data security.

• Eli Lilly is encouraging employees to use AI tools like ChatGPT while taking precautions to protect sensitive information.

• Johnson & Johnson has been using AI for nearly a decade in areas such as surgical assistance, drug discovery, and inventory management. Merck is utilizing generative AI to assist in drafting regulatory documents. Eli Lilly is using AI to research small molecules, generate documentation for clinical trials, and prepare materials for regulatory submissions

💰 OpenAI is investing $12 billion into Coreweave in a five-year deal to acquire cloud services and equity, a strategic move to gain ownership in a key resource provider and reduce reliance on Microsoft.

AvatarOS, founded by virtual influencer creator Isaac Bratzel, secured a $7 million seed round led by M13 to develop a platform for high-quality, personalized AI-powered virtual influencers.

🎮 Microsoft is exploring 3D gaming experiences for Copilot, as indicated by a recent job listing seeking a senior software engineer specializing in 3D rendering engines.

X (formerly Twitter) now allows users to query Grok by mentioning it in replies, expanding access to the AI chatbot on the platform, similar to how Meta is integrating its AI models with its social media platforms.

AI + UI/UX design

Rad AI - Software Engineer Internship

Databricks - Delivery Solutions Architect - Data & AI

Coupang - Staff Machine Learning Engineer, Ads Growth

Zscaler - AI Product Manager

Thank you for staying with us like always! If you are not subscribed, subscribe here for free to get more of these emails in your inbox! Cheers!