Snowflake and NVIDIA Join Forces for Generative AI in the Data Cloud

Learn AI in 5 Minutes a Day

We'll teach you how to save time and earn more with AI. Join 90,000+ free daily readers for trending tools, productivity-boosting prompts, the latest news, and more.

Subscribe

Today:

Snowflake and NVIDIA Team to Help Businesses Harness Their Data for Generative AI in the Data Cloud

Snowflake and NVIDIA have recently teamed up to help businesses become their own AI overlords. Essentially, businesses will be able to make custom AI models using their own data, securely, in the Snowflake cloud.

This tech combo means enterprises can craft advanced AI services such as chatbots, search, and summarization, using their existing data. The head of Snowflake and NVIDIA claim that this partnership will transform the business world by making data more useful.

The 'AI factory' they're building will let businesses utilize their own data to create custom AI models. Even better, customers can build, deploy, and manage these new AI applications to all parts of their business.

This powerful alliance promises to bring its AI magic to various sectors. For example, healthcare could get an AI model that explains complex insurance details, and financial services could get a model that spills the beans about lending opportunities.

Plus, Snowflake plans to host NVIDIA NeMo, a platform for creating AI models, in its Data Cloud. So, it's like getting a guest suite in your house for your very own AI-building software.

READ THE ARTICLE ON NVIDIA

Snowflake launches LLM-driven Document AI

Data cloud company Snowflake is making a significant shift into AI with its new tool, Document AI, announced at their annual conference. This software uses a type of AI known as a large language model to help businesses get useful information from their piles of documents quickly and easily. This is a big step for Snowflake, which previously focused mainly on dealing with structured data, or information that is organized in a specific way, like databases.

Snowflake, based in Montana, also showcased new features like iceberg tables (a way to manage and process large amounts of data), AI-enhanced SQL functions (ways to query or ask questions of databases using AI), and tools to better manage costs at the event.

This shift in focus for Snowflake began after acquiring Applica, a Polish AI firm, in 2022. Applica's tech now powers Document AI. The tool lets users ask for what they need in everyday language, then automatically processes their request to get the required data and insights from a document, such as an invoice or a contract. Once converted, this data can be used for various analytical or AI processes.

READ THE ARTICLE ON VENTUREBEAT

Databricks snaps up MosaicML to build private, custom machine models

Databricks, a data processing platform, is set to buy MosaicML, a company specialized in creating cost-effective, custom AI models, for $1.3 billion. This partnership will make it simpler for businesses to create and use their unique AI models, using their own data.

Databricks sorts and stores data from all over the place into neat little cloud clusters, and MosaicML brings the art supplies for some DIY AI. The benefits are a two-way street. Databricks gets to look cool with some shiny new AI tricks, while MosaicML gets a VIP pass to better data for its custom models. Plus, MosaicML has been showing off with some open-source language models that claim to be as cheap as chips compared to others in the market.

Sure, these models might not have the brains of a GPT-4, but most businesses aren't after a universal genius. With this acquisition, Databricks and MosaicML are well-poised to hunt bigger game in the custom AI market. With a track record of over 10,000 organizations under its belt, Databricks can offer a more appetizing meal deal to potential clients.

READ THE ARTICLE ON THE REGISTER

Unity Unleashes the Power of AI for Creators, Supercharging Content Creation and User Engagement

San Francisco-based Unity, a big cheese in the world of creating real-time 3D content, just launched Unity Sentis and Unity Muse. These are two new AI-based tools to help folks who make games and experiences work faster and make their 3D projects more engaging.

Unity Sentis is an engine that deploys AI models in any Unity project. Basically, it helps make games more interactive and lively. It has features like Muse Chat which allows creators to get help and info fast just by typing into a chat box. More features are expected in the coming weeks.

Unity also launched a special AI marketplace where creators can find third-party tools and stuff to speed up their work. Both Unity Sentis and Unity Muse are currently being tested in a closed beta, but they're expected to be available for everyone later this year.

READ THE ARTICLE ON YAHOO! FINANCE

Tempo’s new take on AI personal training adds 3D body scans and dynamic reps

Tempo, an at-home fitness tech company, is set to revolutionize strength training with a major update. This upgrade includes 3D body scans and AI-powered fitness classes that adjust based on your performance and offer immediate feedback on form. This new approach enables a tailor-made fitness experience, adjusting the difficulty, intensity, weights, rep targets and rest periods according to your specific fitness data and readiness.

The new feature set utilizes data gathered from wearables like the Apple Watch and can interact with platforms like Strava or Garmin for comprehensive fitness tracking. The tech assesses how hard you've worked during a set, offering suggestions to increase reps or weights if you've got more gas in the tank, and automatically increasing recovery time if your heart rate is too high.

Additionally, Tempo introduces the option for full-body scans. This feature uses the iPhone's cameras to create a 3D avatar of your body, estimating your body fat and muscle mass based on weight, height, and gender data. Users can opt to keep a visual record of their progress or delete the image, keeping only the numerical data.

READ THE ARTICLE ON THE VERGE

Thomson Reuters buys Casetext, an AI legal tech startup, for $650M in cash

Thomson Reuters, a multinational mass media firm, has decided to purchase Casetext, a company creating artificial intelligence (AI) tools for law practices, in a deal worth $650 million. The plan is to wrap this up in the latter half of 2023, but that depends on some routine checks and balances.

Casetext is known for using AI to build workflows and tools for legal professionals. Their top product, CoCounsel, uses AI to check documents, aid with legal research, prepare depositions, and scrutinize contracts. This platform relies on OpenAI’s GPT-4 language model to work its magic.

Thomson Reuters' decision to buy Casetext falls in line with their long-term goal of adding AI to their main areas of business — law, taxes, accounting, and news. They have even declared their intention to spend about $100 million each year on AI, and to commit $10 billion for buying other companies by 2025. Many of these purchases will likely be AI-focused.

READ THE ARTICLE ON TECHCRUNCH

Loora, a generative AI app that uses an audio interface to help users learn English, raises $9.25M

Loora, an AI-based language learning startup, has emerged from stealth with $9.25 million in seed funding. The app specializes in enhancing English conversation skills through voice-based interactions with its AI tutor, helping to improve English language proficiency and accents across diverse topics. The startup is unique in its audio-focused approach, a departure from the primarily text-based AI learning apps prevalent in the market.

The funding comes primarily from three early-stage investment companies: Emerge, Two Lanterns Venture Partners, and Kaedan Capital. Based in Tel Aviv, Loora has been operating under the radar and already has thousands of paying customers, both everyday consumers and professionals needing to improve their English.

Loora uses advances in AI, specifically generative AI used by ChatGPT and Midjourney, to create an English tutor that interacts through conversation, helping improve spoken English. It initially built its own language models but now uses a variety from different sources, training them on its own data.

READ THE ARTICLE ON TECHCRUNCH

AI predicts hit songs based on listeners’ heartbeats

Scientists have taught a computer system to predict chart-topping tunes with impressive precision, by simply checking out how your heart dances to the beat, without any need to dissect the song itself.

In earlier work, scientist Paul Zak from Claremont Graduate University, noticed that tiny shifts in heartbeats can hint at brain activity related to focus and emotion. After chatting with a music streaming service about their struggle to recommend new music, Zak thought measuring these changes in the brain, which he calls 'Immersion', could be the solution.

To test this, the streaming service provided Zak's team with 24 newly released songs, split evenly between hits and flops. They then had 33 volunteers listen to these tunes while wearing heartbeat monitors. Zak's team found that the hits sparked higher "Immersion" scores in the listeners' brains.

Since 24 songs isn't enough to train a computer system, they created a simulated dataset of 10,000 brain reactions to songs. They trained the computer system on half of this data to sort songs into hits or flops. When tested, it nailed the classification 97% of the time, which is unheard of in the industry.

READ THE ARTICLE ON FREETHINK

Superforecasting with AI promises the best of all future worlds

Tetlock brought together a bunch of really smart people (superforecasters) from different fields and got them predicting things like election results and stock market moves. Turns out they did a great job, outperforming even security analysts who had classified intel.

But these superforecasters hit a snag when predicting outcomes in geopolitical negotiations. Tulchinsky thinks AI could help there. AI can chew through data faster than any human, allowing us to ask more questions and test theories more thoroughly.

Tulchinsky's method for reliable forecasting is to use both AI and a team of superforecasters to test out theories. It ain't perfect, but it's the best we've got right now. The big takeaway is that the best use of AI is to give human smarts a boost, not to try and take over.

READ THE ARTICLE ON FINANCIAL TIMES

IBM’s HR team saved 12,000 hours in 18 months after using A.I. to automate 280 tasks

IBM's HR team has utilized AI automation to increase efficiency and cut manual labor hours by a whopping 12,000 in the last 18 months. The Watson AI system, originally demonstrated on Jeopardy!, now helps employees with mundane tasks and queries like vacation policies. IBM also uses AI to automatically sift through performance data to make suggestions for raises or promotions, freeing managers to focus on coaching employees and career development.

IBM's Chief Human Resources Officer, Nickle LaMoreaux, emphasizes the importance of reskilling and training in order to maintain an indispensable workforce. AI-related skills are being sought after in job candidates, often acquired through nontraditional methods like community colleges or online boot camps.

READ THE ARTICLE ON FORTUNE

How AI is helping to shrink waiting times for NHS cancer patients

At Addenbrooke’s Hospital in Cambridge, a new AI system, "OSAIRIS," is shortening the waiting times for cancer patients requiring radiotherapy. Developed by and for the NHS, OSAIRIS speeds up the preparation of scans, reducing the time patients have to wait from referral to treatment start.

The system helps medical specialists to plan treatments roughly two and a half times faster than they could on their own. It's initially being used for prostate and head and neck cancers, but has potential across various cancer types within the NHS. OSAIRIS lightens the workload of doctors by doing the time-consuming task of outlining healthy organs on scans ("segmentation") before radiotherapy, which can take between 20 minutes to three hours per patient. This lets doctors focus more on treatment planning.

The AI tech was developed as part of a collaboration between Dr. Raj Jena, an oncologist at CUH, and Microsoft Research on Project InnerEye. With the aid of a £500,000 grant from the NHS AI Lab, Dr. Jena's team created OSAIRIS using Project InnerEye's open-source software technology and Azure Machine Learning. Rigorous testing confirmed that OSAIRIS is safe for daily use in caring for radiotherapy patients.

READ THE ARTICLE ON MICROSOFT