Google's New AI Tool Can Do Research for You

Hello, AI Enthusiasts!

Welcome to FavTutor’s AI Recap! We’ve gathered all the latest and important AI developments for the past 24 hours in one place, just for you.

In Today’s Newsletter: 😀 

• Google's New AI Tool Can Do Research for You

• Meta Launches AI-Generated Video Watermarking Tool

• OpenAI's o1 relies on trial-and-error and informal reasoning

Google
🔍️ Google's New AI Tool Can Do Research for You

Google has launched an AI tool that conducts research on your behalf and churns out detailed, easy-to-understand reports so you don't have to. The new Deep Research tool runs on the Gemini 1.5 Pro model and is available on desktop and mobile via a web browser.

Insights for you:

• Deep Research uses AI to explore complex topics and provide you with findings in a comprehensive, easy-to-read report.

• With this tool, you can ask Gemini to research a particular topic. In response, the tool will develop a multi-step research plan. It will browse the web for different takes and refine its analysis to create a comprehensive report, including links to relevant sources.

• You can also ask follow-up questions or ask for specific edits like adding new sections of information.

Streamline your development process with Pinata’s easy File API

• Easy file uploads and retrieval in minutes

• No complex setup or infrastructure needed

• Focus on building, not configurations

Try today!

Meta
©️ Meta Launches AI-Generated Video Watermarking Tool

Meta’s Video Seal adds a hidden watermark for AI-generated video, which cannot be modified by blurring or cropping the video. In addition to a watermark, Video Seal can add a hidden message to videos that can later be uncovered to determine their origins

Insights for you:

• Meta reveals a new Video Seal tool to apply imperceptible watermarks to AI-generated video clips.

• This open-source tool joins Meta’s other watermarking tools, Watermark Anything and Audio Seal.

• Meta offers this Video Seal openly under a permissive license, along with research papers, training code, and inference code.

AI Research
📜 OpenAI's o1 relies on trial-and-error and informal reasoning

A new study comparing public and private datasets shows how OpenAI's o1 model approaches mathematical problem-solving. The research aims to determine whether the AI uses actual logical reasoning or simply recalls memorized solutions.

Insights for you:

• A new study examines OpenAI's o1-mini model and shows that it is unlikely to rely on memorization for its solutions.

• This was demonstrated by comparing public Maths Olympiad tasks with private training tasks. The model achieved similar results for both data sets: around 70% accuracy for search tasks and 21% for calculation tasks.

• However, the system shows weaknesses when formulating detailed mathematical proofs and uses a trial-and-error approach instead. The researchers observed that o1-mini exhibits human-like mathematical intuition and can recognize important intermediate steps, even if the formal elaboration remains incomplete.