Google Meet will now create your To-Do List

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 Meet will now create your To-Do List

• Evo-2 is the biggest-ever Biology AI Model

• New S* framework helps AI models write better code

Google
📜 Google Meet will now create your To-Do List

Google has announced a new AI feature for its video conferencing platform: a new “suggested next steps” section will help it determine the next steps discussed during the meeting. These follow-ups will automatically be identified and organized within the meeting notes document.

Insights for you:

• Google Meet is rolling out a new “next steps” feature that uses AI to identify and organize follow-up tasks discussed during meetings.

• The update builds on Google Meet’s “Take notes for me” feature, which captures key discussion points during meetings.

• The feature will be available to Google Workspace users with Business and Enterprise plans.

Evo
🧬 Evo-2 is the biggest-ever Biology AI Model

A research team has developed Evo 2, which they describe as the largest AI model yet created for biological applications. The model learns to precisely predict how genetic variants affect function just by analyzing DNA sequences, without requiring additional task-specific training.

Insights for you:

• Researchers have developed Evo 2, the most comprehensive AI model for biological applications to date, trained on a genomic atlas of 9.3 trillion DNA base pairs.

• Evo 2 can independently learn a large number of biological features, generate complete genomes, and design complex epigenomic structures such as chromatin accessibility in a controlled manner using inference-time search.

• The team developed two versions of Evo 2, featuring 7 and 40 billion parameters respectively. Both can process sequence contexts up to 1 million base pairs long.

AI Research
👨‍💻 New S* framework helps AI models write better code

Researchers at the University of California, Berkeley have created a framework called S* that improves how AI language models generate code. However, it's currently optimized only for programming competition and hasn't been tested on more complex software engineering.

Insights for you:

• Researchers have developed the S* framework, which improves the performance of AI language models in code generation by combining parallel and sequential scaling approaches.

• S* uses a language model to generate test inputs that are particularly suitable for detecting differences between different program solutions. The actual results of these tests are then used to select the best solution.

• In the evaluation of 12 different language models, S* consistently improved performance. Small models with S* even outperformed large reasoning models without S*.