Future Blueprint
Posts
💡 Bill Gates loves AI

💡 Bill Gates loves AI

Plus: Latest News, Tools, and Advancements

Read time: 3 minutes

Welcome back, AI buffs! It's time to uncover the latest AI innovations and research together.

In Today’s AI Report:

💰 Bill Gates on AI: The billionaires’ thoughts
🔋 New AI Model: Save energy with this AI
🤖 MultiModal-GPT: Chat GPT crusher?
💎4 AI Tools 4 You: Handpicked just 4 you

💰 Bill Gates on AI

In a recent interview with Bill Gates, he shared his opinion on AI stating that it has the potential to be revolutionary. His views on AI go against the views of Geoffrey Hinton, ‘The Godfather of AI’ who recently backed away from AI advancements.

Gates realizes the possibilities of generative AI, especially in healthcare and education. In healthcare, there have been multiple breakthroughs, some as early as last week. You can read about the breakthrough in my previous newsletter.

He has been impressed with the power behind GPT-4 but is aware that, without precautions, AI could be harmful in the wrong hands.

“I see the AI, used properly, is providing a lot of benefits that, no, I wouldn't want to throw those away.”

Bill Gates

🔋 New AI Model

Researchers from the Dutch National Research Institute for Mathematics and Computer Science have made strides in a new AI model. It works faster while expending less energy.

It is designed to work on VR headsets and smartphones. The study showcases how brain-like neurons combined with innovative learning methods can create large scale training models with a small amount of energy.

The new learning method allows the AI to learn directly from data, which makes it more efficient and powerful. This could lead to smarter, low-power devices in the future.

🤖 MultiModal-GPT

Introducing MultiModal-GPT: “A vision and language model for dialogue with humans. What does this mean? In AI multimodal means that an AI modal is capable of understanding different data like text, images, or audio.

The new model is capable of generating captions, counting objects, and answering general inquiries.

It was created by constructing instruction templates that incorporate vision and language data through a system called “LoRA.” The Low-rank Adapter (LoRA) is a technique used to incorporate vision and language data into the MultiModal-GPT model.

You can try out the demo here: (link)

Read the full paper here: (link)

💎4 AI Tools 4 You

Dante AI: All-in-One AI tool (link)
SciSpace Copilot: Chrome extension that summarizes papers in real time (link)
Uizard: Turn simple sketches into a website (link)
Chat GPT: AI tool trained to respond like a human (link)