- The Daily Update
- Posts
- Microsoft's AI for mobile devices
Microsoft's AI for mobile devices
Spotlight: Google unveils new healthcare tools
Jack Shields
December 14, 2023
Welcome back!
ChatGPT might be getting lazier during the holiday season, but Big Tech certainly isn’t. Microsoft and Google just made big announcements and I’ve got the details for you here.
Let’s go.
In today’s Daily Update:
🗞️ Microsoft releases new small language model
🤖 Visual Electric — AI image generation for designers
📸 Google unveils family of healthcare-focused AI models
🚨 AI Roundup: Four quick hits
Read time: 2 minutes
TOP STORY
🗞️ Microsoft releases new small language model
Source: Microsoft
Microsoft Research just released Phi-2, a powerful small language model (SLM) capable of running on a laptop or mobile device.
The details:
Phi-2 is a text-to-text model that demonstrates impressive capabilities in reasoning, language, math and coding.
According to Microsoft, Phi-2 matches or outperforms models up to 25x larger due to new training innovations.
Microsoft is currently licensing the model for non-commercial research purposes.
The relevance: As tech giants race to build the world’s most powerful AI, Microsoft has subtly introduced a more efficient way to run AI models locally on laptops and smartphones. Phi-2 and future SLMs could integrate AI capabilities into the devices you use everyday.
AI TOOL OF THE DAY
🤖 Visual Electric — AI image generation for designers
Generated by Visual Electric
Visual Electric is an AI image generator specifically built for designers. It’s easy to use and offers a variety of ways to fine-tune your ideas. Here’s how it works:
Step 1: Create a new canvas and add a text prompt (ex. “a bunny drinking beer on a frozen lake under the northern lights in a winter wonderland”).
Step 2: Select a mood and aspect ratio (ex. “photography, 16x9”).
Step 3: Generate images.
Step 4: Fine-tune the results (remix, upscale, touch up or make variations of an image).
This is an impressive image generator that gives users much more control over its outputs than other text-to-image models. I highly recommend giving this tool a try.
BUSINESS SPOTLIGHT
📸 Google unveils family of healthcare-focused AI models
Generated by Visual Electric
Yesterday Google announced MedLM, a family of AI models fine-tuned for the healthcare industry.
What you should know:
MedLM is based on Med-PaLM 2, a Google-developed model that performs at an “expert level” on dozens of medical exam questions.
Google hopes to enable healthcare workers to offload tasks to generative AI, or at least use AI to assist in completing their tasks.
There are two MedLM models available currently: a larger model designed for complex tasks and a smaller one for “scaling across tasks.”
Research has shown that generative AI isn’t particularly reliable in healthcare, but Google asserts that it’s being extremely cautious in releasing AI healthcare tools.
Crucial quote: “We’re focused on enabling professionals with a safe and responsible use of this technology,” VP of engineering and research at Google Yossi Matias said. “And we’re committed to not only helping others advance healthcare, but also making sure that these benefits are available to everyone.”
MORE TRENDING NEWS
🚨 AI Roundup: Four quick hits
Source: The Washington Post
Fun Feature: Washington Post columnist uses AI to make a Christmas card.
Smart Spreadsheets: Anthropic brings its Claude LLM to Google Sheets.
Breaking Bread: OpenAI inks deal with Business Insider and Politico publisher to license news articles for AI training.
Synthetic Socials: Snapchat now lets Plus users share AI-generated snaps.
THAT’S ALL FOR TODAY