- The Daily Update
- Posts
- OpenAI partners with Moderna
OpenAI partners with Moderna
Spotlight: Microsoft pursues small language models
Jack Shields
April 29, 2024
Welcome back!
We’re covering a lot of ground today. AI-powered medicine, video generation and a new approach to language models are all on deck. Let’s go.
In today’s Daily Update:
🗞️ OpenAI partners with Moderna to advance medicine
🤖 Adobe develops tool for upscaling blurry videos
📸 Microsoft launches Phi-3 small language models
🚨 AI Roundup: Four quick hits
Read time: 2 minutes
TOP STORY
🗞️ OpenAI partners with Moderna to advance medicine
DALL-E 3
OpenAI and pharmaceutical giant Moderna just announced their ongoing collaboration to advance mRNA medicine.
What you should know:
Since adopting ChatGPT Enterprise, Moderna has deployed more than 750 GPTs to drive automation and productivity.
These custom GPTs partially automate tasks like prescribing optimal drug doses to patients.
OpenAI says each “weekly active user” averages 120 ChatGPT conversations per week.
Moderna CEO Stéphane Bancel says OpenAI’s tech enables his team of a few thousand employees to perform like 100,000.
Why it matters: mRNA technology has surged in popularity since Moderna used it to develop one of the earliest COVID-19 vaccines. Beyond accelerating the development of new treatments for diseases, Moderna’s AI adoption is a compelling case study on how generative AI can boost the productivity of entire organizations.
AI TOOL OF THE DAY
🤖 Adobe develops tool for upscaling blurry videos
Source: Adobe
Adobe researchers have developed a generative AI model that can upscale blurry videos up to 8x their original resolution.
The details:
Common GAN (Generative Adversarial Network) approaches are effective for upscaling still images but struggle with videos.
VideoGigaGAN aims to fine-tune details without adding any mistakes to video footage.
Adobe says their new model upscales videos to a higher resolution without adding flickering or other unwanted elements.
The tool is currently in research preview, so it may not be released to the public.
Why it matters: This is another important development for AI video generation. After announcing generative AI features for Premiere Pro last week, Adobe seems to be leading the charge toward fully AI-generated short films.
BUSINESS SPOTLIGHT
📸 Microsoft launches Phi-3 small language models
Source: Microsoft
Microsoft announced the launch of its open-source family of Phi-3 small language models last week.
Key points:
Small language models are designed to perform well on simpler tasks, and are easier for organizations to fine-tune to meet specific needs.
They also enable organizations to build applications that run locally on a device instead of through the cloud.
Phi-3-mini, the first of these models to be released publicly, reportedly performs better than models twice its size.
Microsoft says additional Phi-3 models are coming soon.
Why it matters: While the tech giants race to build the world’s biggest and most powerful LLM, there’s a lot of upside to efficient small language models. Smaller models like Phi-3 could accelerate the development of on-device AI and offer affordable AI solutions to small businesses.
MORE TRENDING NEWS
🚨 AI Roundup: Four quick hits
DALL-E 3
Apple releases OpenELM family of language models optimized for efficiency (similar to Phi-3 above).
Baltimore gym teacher accused of using AI voice clone to try to get high school principal fired.
Drake takes down song featuring AI Tupac after legal action was threatened.
Google rolls out AI conversation practice for English language learners.
THAT’S ALL FOR TODAY