AI for the Good of Humanity?

So this week, Google launched their Med-Gemini, a variant of Gemini model that beats even GPT-4 in medical based use cases. While Sam Altman does call GPT-4 dumb, surpassing this model in smartness is a feat very few models hold.

A few days back we read how AI Model was used to detect a disease that doctors and traditional machinery and models could not catch.

Some AI Tools are already making headlines on breast cancer detection.

I honesly love these sorts of use cases, it benefits the humanity as a whole. Imagine a world where we build AGI, and AGI instead of taking jobs, assists in endless research for healthcare to improve quality of human life overall.

Google took a huge step forward in this direction last week. Here’s the scoop:

Google launches Med-Gemini

Google has unveiled a new family of AI models called Med-Gemini, designed specifically for healthcare applications. These models are built upon Google's previous Gemini models and are aimed at transforming various aspects of the medical industry. The key capabilities of the Med-Gemini models include assisting clinicians in medical decision-making, helping interpret complex lab and imaging data, and enabling early disease detection. Additionally, the models can provide personalized health coaching based on Fitbit data, offering tailored guidance and recommendations to individuals. The Med-Gemini models have achieved significant improvements in accuracy and scalability, with the larger Med-Gemini-L 1.0 model outperforming its predecessor Med-PaLM 2 by 4.5%. The medium-sized Med-Gemini-M 1.0 model offers more scalable capabilities.

Announcement Tweet shows how the model can help with early diagnosis and detection of diseases.

https://twitter.com/i/status/1785117450528264216

Amazon launches Q: An AI Powered Assistant

Amazon's Q AI, a generative AI-powered assistant, is a transformative tool for businesses and developers, now generally available. It offers advanced capabilities like generating accurate code, testing, debugging, and multi-step planning, enhancing software development and leveraging internal data. With a focus on boosting workforce productivity, Amazon Q empowers employees to streamline tasks, build custom AI-powered apps, and access business insights efficiently. Its integration with over 40 data sources and enterprise systems ensures seamless access to information, enabling users to be more creative, data-driven, and efficient.

Additionally, Amazon Q's security features prioritize data protection, making it a reliable solution for accelerating decision-making and innovation within organizations.

Apple's plan for WWDC 24

Apple has recently opened a secret AI lab called the Zurich Vision Lab, which will focus on advanced AI research, specifically for building the Vision Pro and its applications. Over the past two years, Apple has acquired more than 30 AI startups and recently released OpenELM, their own AI models fine-tuned for mobile devices.

These developments come just a month before the Apple Worldwide Developers Conference (WWDC) 2024, scheduled for June 10-14. Despite Apple's lack of significant AI progress in the past two years, the company has been focusing on acquisitions, with AI-related actions starting only in March.

It appears that Apple may be rushing to launch AI-specific announcements at WWDC, possibly due to being late to their original plans. There are also rumors about Apple integrating Google's Gemini between iPhones, but these remain unconfirmed. The upcoming event is expected to be exciting, with anticipation building around Apple's AI-related revelations.

AI Products launched this week 

• Otterly.ai - Otterly.AI is the new way in content & brand monitoring for AI-powered Search Experiences. Monitor your Brand, Content, Competitors or any other Prompt.

• Outerbase 2.0 - Outerbase is the AI powered interface for your database. Allowing you and your team to view, query, and visualize your data using SQL or natural language. 

• Braudit - Braudit enables you to document your career journey as a story with chapters, discover insights about your experiences and personality and get data-driven advice in seconds.

• Claude for iOS - Claude, the AI assistant built by Anthropic, is now available on your iPhone.

• Mindtrip - Mindtrip is a travel platform that combines conversational AI with a proprietary travel knowledge base to deliver personalized travel experiences that are accurate, actionable, and all in one place.

• Noodle4 AI- We can reduce content review up to 70%, by providing brands/ agencies with a plug & play AI platform. Our custom built LLM allows you to upload your brief document and all associated content and Noodle4 will analyse the content in a simple UX.

This week in AI

• Altman called GPT-4 “the Dumbest AI Model” and most of us are paying for it 😊

• Turn Your Selfie into a Dance Video -  Diffuse by Higgsfield uses AI to create personalized dance videos from selfies. It's user-friendly, offers free trials, and is popular among content creators for its unique features, allowing users to showcase their style and personality through entertaining videos.

• OctoAI – OctoAI's OctoStack integrates generative AI into enterprises' data infrastructure securely. It's a turnkey GenAI stack for their cloud, enabling AI-powered apps that bring AI to their data, not vice versa. OctoStack seamlessly integrates with Snowflake, unlocking new data workflows using Retrieval Augmented Generation to increase trust in GenAI solutions.

• SK Hynix says HBM chips almost sold out for 2025- SK Hynix, a major Nvidia supplier, is almost sold out of high-bandwidth memory (HBM) chips for 2025. The demand surge, driven by AI services expansion, highlights the growing need for potent AI processors. Samples of their latest 12-layer HBM3E chip will be sent out in May, signaling a promising future for AI industry advancement.

• Google's RealFill  - It is an AI-powered image completion model that accurately fills in missing parts of images using reference images. It represents a significant advancement in AI-driven image generation, with trademark filings indicating Google's intent to commercialize this innovative technology for user-centric applications, potentially revolutionizing image completion and generation.

• Chinese AI Robot:  The AstriBot S1 is an autonomous AI robot known for its remarkable speed and accuracy. It showcases impressive capabilities in sorting objects quickly and accurately without human control, demonstrating autonomous functionality. The robot's design allows it to identify and place objects back in their correct positions swiftly, making it stand out in terms of efficiency and precision compared to other robots in similar tasks. Additionally, the AstriBot S1 has demonstrated unique features like jumping, showcasing a blend of agility and advanced robotic capabilities.