- Starter AI
- Posts
- Meta’s SAM 2, Voice Mode is coming, Baidu’s approach to hallucinations
Meta’s SAM 2, Voice Mode is coming, Baidu’s approach to hallucinations
The flood of innovation doesn't stop.
Sthefania
July 31, 2024
In partnership with
Hello, Starters!
July has been a very busy month in tech. Companies have been continuously sharing their latest advancements, and it seems that they don't plan to stop for now. We're considering calling this the summer of AI.
Here’s what you’ll find today:
Meta releases SAM 2
OpenAI starts to rollout “Voice Mode”
Baidu’s research on hallucinations
Runway’s new Image to Video model
Getty and Shutterstock’s AI work with Nvidia
And more.
🤖 Meta releases SAM 2 (1 min)
Meta continues to thrill the AI community, this time by presenting a new generation of its Segment Anything Model. SAM 2 is a unified model that allows seamless real-time object segmentation in images and videos. As expected, they're also sharing the code and weights for the model under an Apache 2.0 licence, promoting once again openness in development.
Segmentation is based on identifying the pixels corresponding to a specific object of interest in an image, mostly used in computer vision. Meta points out that SAM 2 will be a great aid to bring innovations in this field, as well as allow users to create new video effects, among other applications.
🗣️ OpenAI starts to rollout “Voice Mode” (2 min)
A number of ups and downs have followed OpenAI's announcement of its advanced “Voice Mode” feature, which is one of the most expected this year. After delaying its release last month to improve its safety, they've just started giving access to selected users of ChatGPT Plus.
Advanced Voice Mode is scheduled for a gradual release that OpenAI anticipates doing this fall. However, the demo that left most users in awe will not be fully blown out yet, like the video or screen-sharing capabilities, which will be coming at a later date.
🧠 Baidu’s research on hallucinations (4 min)
Hallucinations are still one of the main issues most models face, and finding solutions to them has been a constant focus for companies that develop AI technologies. Baidu is one of them, and they've recently shared a paper that contains its research on how a "self-reasoning" framework can help AI systems stop giving incorrect information.
This method consists of allowing the models to evaluate their own knowledge by focusing on their reasoning trajectories and putting them through a series of processes that improve their decision-making and enable the system to critically analyse its outputs.
FREE AI & ChatGPT Masterclass to automate 50% of your workflow
More than 300 Million people use AI across the globe, but just the top 1% know the right ones for the right use-cases.
Join this free masterclass on AI tools that will teach you the 25 most useful AI tools on the internet – that too for $0 (they have 100 free seats only!)
This masterclass will teach you how to:
Build business strategies & solve problems like a pro
Write content for emails, socials & more in minutes
Build AI assistants & custom bots in minutes
Research 10x faster, do more in less time & make your life easier
You’ll wish you knew about this FREE AI masterclass sooner 😉
🎥After announcing Gen-3 Alpha in June, Runway is now adding a new Image to Video feature to the tool which empowers users with even more creative freedom. Images can be used on their own or with additional prompts to generate videos from 5 to 10 seconds.
💥Both Shutterstock and Getty Images are diving into generative AI with the help of Nvidia. From Shutterstock's side, its latest service allows users to create 3D assets through images or text prompts, and Getty has made enhancements to their image generator.
⚡️Quick links
What did you think of today's newsletter? |