One step closer to AGI

Hey there,

"AI" is currently the most impactful term in the US stock market, and companies are eagerly projecting their future growth by leveraging it. 

Palantir is up 32% this week after stating “the demand for AI is Unprecedented” (read more)

In today’s edition:

📰 News: Multi-Modal AI by Meta is one step closer to AGI

🛠️ Tools: Chatbot to answer questions about any PDF document.

Reading time is 3 minutes. Let’s dive in 🤿

One step closer to AGI

What: Human ability to use multiple senses together makes us unique. For example, you could probably visualize a flower just by its smell.

But the current generation of AI models is limited to 2 "dimensions". For instance, the state of the art GPT-4 AI model is limited to text and images.

Meta released a new open-source AI model called ImageBind, that binds 6 different dimensions together! It can understand text, audio, image/video, 3D shape, temperature, and movement and use one or more of them to retrieve another.

Why it matters: ImageBind is a significant step towards the development of general artificial intelligence, enabling it to become aware of its surroundings.

For instance, say you want to visualize a rain forest. Current AI models can generate a high resolution image for rain forest.

But ImageBind takes this to whole new level! It can also generate ambient sounds like rain and leaves hustling, along with depth, thermal imaging and correct motion dynamics of objects in the image.

What’s next: ImageBind can greatly simplify creative workflows. Example, in video editing, the AI could isolate individual sounds/objects using plain text prompts or animate static images with perfect natural sounds to create immersive experiences.
Plus, Meta has been open-sourcing the latest AI research in an effort to build a strong developer base.

New Tools
Leverage AI to grow

• PDF.ai: Ask questions, get summaries and find information from any document, from legal agreements to financial reports.(link)

• Inbox Narrator: Summarizes your emails and delivers them in an impressive human-like voice. (link)

• VizGPT: Designed to to generate visualization from dataset using natural language along with feature to edit and explore your visualization using chat context. (link)

• Flippit: A no-code tool to craft smart and interactive AI avatars for virtual experiences. (link)