Meta AI+SAM=❤️🔥. Hey Doc, AI & chill? IBM intros AI golf commentator. Nvidia's reign...over?
GM, and happy Thursday! Welcome back to another AIcyclopedia newsletter.
BIG NEWS: The AIcyclopedia community is about to hit 3,000 members. With the help of Aquaman, we want to take a moment and say THANK YOU for joining us each week and helping us hit another milestone!
Now, here’s what we’ve got for ya today:
Meta AI launches SAM - A revolutionary model for segmenting anything
Overwhelmed doctors get a helping hand from AI
AI takes on a new role: Sports Commentator
Google and Nvidia Face Off: The Battle for AI Dominance
Top AI news you need to know: Part ☝🏾
1/ Meta AI launches SAM - A revolutionary model for segmenting anything
Meta, it’s about time.
[The News] Meta AI announced the release of the Segment Anything Model project (SAM) on Wednesday.
The dataset used is said to be the largest of its kind.
[The AI] The SAM model allows users to identify, pick out, and separate individual objects (or segments) in a given image using clicks and text prompts.
According to Reuters, “In one demonstration, writing the word "cat" prompted the tool to draw boxes around each of several cats in a photo.”
This is possible as the AI is trained on over 1 billion segmentation masks collected from 11 million licensed images.
(Read the full Meta AI post here.)
[Why this matters] Let’s say you wanted to create a similar model of your own.
There’d be a few ingredients you’d need: time, money, experts, & a LOT of data.
But now, AI practitioners will no longer need to collect their own segmentation data.
Because Meta is serving up SAM for public use, this will “enable a wide range of applications and encourage more research into foundation models for computer vision.”
Looking forward… Meta states that “SAM can become a powerful component in domains such as AR/VR, content creation, scientific domains, and more general AI systems.”
There are currently 2 releases:
The Segment Anything Model (SAM)
Segment Anything 1-Billion mask dataset (SA-1B)
SA-1B is only available for research purposes, but luckily for the public…
Anyone can access the SAM demo. (Test it out here.)
****
2/ Overwhelmed doctors get a helping hand from AI
“Hey Doc, can I schedule an appointment with you today?”
"Ah, sorry. Not today. I’m drowning in paperwork.”
[The News] According to research published in 2016, doctors spend almost half of their time on administrative tasks and paperwork versus face-to-face time with their patients.
One company was aware of the issue and decided to use artificial intelligence to help solve it.
[The AI] Glass Health, a San Francisco-based startup, is an AI tool for doctors to help minimize the paperwork burden they face every day.
To give you some insight, the AI is similar to ChatGPT but is trained using data from a virtual medical textbook written by medical professionals.
Derek Paul, the co-founder, told NPR, “We're working on doctors being able to put in a one-liner, a patient summary, and for us to be able to generate the first draft of a clinical plan for that doctor. So what tests they would order and what treatments they would order."
[Why this matters] The biggest concern for doctors is burnout and not allocating enough time for patients as paperwork piles up.
With AI like Glass Health, doctors can focus on helping patients and increase their own quality of life and work-related efficiency.
Looking forward… Right now, patient bias and wrong information are the biggest problems with using AI models to help doctors.
Because the AI relies on “written diagnoses and treatment plans,” working with a diverse set of doctors around the world will help reduce bias and misinformation.
According to Paul, Glass Health hit 4,800 users in February as more and more doctors became interested in incorporating AI to reduce manual paperwork.
Want to sponsor a future newsletter and share your business with our growing list of followers?
Go ahead and send an email to [daniel@aicyclopedia.com].
This space could be yours…🥳
Top AI news you need to know: Part ✌🏼
3/ AI takes on a new role: Sports Commentator
Hey golfers, AI live commentary is takin’ over.
[The News] According to Sports Illustrated, the 2023 Masters event is partnering with tech giant IBM to provide commentary using AI.
[The AI] For more than 20,000 clips on the Masters app, the AI being used is Watson AI, IBM’s cloud-based API chatbot service.
In short, it was trained to understand “golf terminology and analyze the swings in each video.”
Here’s the result:
Looking forward… Will sports commentators be out of a job?
As of right now, it doesn’t look like it. It works for an app because it streamlines mass content creation and provides listeners with quick and immediate context for a clip they are watching.
Scanning the Twitter-verse, people were quick to point out that AI still lacks the "expression and color" that many of us enjoy from a live sporting event’s commentary.
****
4/ Google and Nvidia Face Off: The Battle for AI Dominance
Google appeared to lag behind the rest of the world when it released Bard months after ChatGPT.
But now, Google is wide-eyed and making moves to get ahead in the AI race.
[The News] In case you didn’t know, Nvidia captures over 90% of the AI model training and deployment space with its $10,000 Nvidia A100 chip.
But Google announced Wednesday that it’s introducing its own AI “supercomputer” that is “faster and more efficient” than Nvidia.
It is called the Tensor Processing Unit v4 or TPU v4.
[The AI] According to Google’s researchers, the 4th and newest iteration of their TPU is “1.2 - 1.7x faster and uses 1.3x-1.9x less power than the Nvidia A100.”
Looking forward… Google currently has the chance of capturing a larger portion of the AI chip market.
But the big question is: Can they catch up to Nvidia?
The TPU v4 results were compared against the Nvidia A100 chip, not the newest H100.
So Google, you better cross those fingers. 🤞🏼🤞🏼
Your AI news break 🥳
AI News Tid Bits…
👉🏻 Boston University embraces AI and implements new AI policy to help student with assignments. Right now, this is simply a baseline policy and will evolve alongside AI.
👉🏻 The first AI Fashion Week in April will feature designers who’ve embraced AI as a tool in their design process. The question is: Will AI fashion stick around, or is this simply a fad?
👉🏻 Levi’s partners with AI digital fashion studio to introduce AI-generated models. The reason? To “increase diversity.”
Try this AI tool 🪄 : Fathom
According to Product Hunt, Fathom records, transcribes, and replaces note taking with a click of a button on your Zoom calls, so you can simply focus on the conversation at hand.
The best part?
Fathom is FREE to use with NO usage limitations whatsoever. Seriously.
Here’s what people are saying:
I don't know how this is free but I absolutely love it. The AI summary is what I've been looking for. I no longer have to take notes OR press any buttons. It's all done for me and summarized perfectly, even when talking about very niche topics.
Love this tool! It saves me so much time! My coaching clients LOVE to receive their session recordings with all of the notes and highlights. They are really impressed by it over the standard Zoom recordings.
Blown away with the AI note-taking, it was concise and accurate. This is an incredible product, it just keeps getting better and better. Highly recommend it!!
Here’s a great review and tutorial from Feisworld Media:
That’s it for today!
If you liked the newsletter, give it a ❤️ and, as always, feel free to sound off in the comment section below 👇🏾
Thank you for choosing us as your AI news source, and don’t forget to follow us on Twitter!
See you Sunday 👋🏻