🤖 Microsoft’s New “Recall” Feature: Pivotal or Privacy Nightmare?

Subscribe | Contact | Meet The Team

Welcome back AI enthusiasts!

In today’s AI Report:

• 🔐Microsoft’s New “Recall” Feature: Pivotal or Privacy Nightmare?

• 🗺Anthropic Maps the Minds of LLMs

• 🎓Google’s Free AI Certification Courses

• 🛠5 Trending Tools

• 💰Venture Capital Updates

• 💼Who’s Hiring?

Read Time: 3 minutes

🗞RECENT NEWS

MICROSOFT

🔐Microsoft’s New “Recall” Feature: Pivotal or Privacy Nightmare?

At Microsoft Build 2024, the company introduced various new AI features such as “Copilot+ Recall,” “Team Copilot,” and a “Phi-3-vision” multimodal model.

Key Details:

• Microsoft CEO Satya Nadella discussed the new Windows 11 “Copilot+ Recall” feature, which tracks everything you see and do on your device.

• “Copilot+ Recall” constantly takes screenshots of what’s on your screen. Then, it leverages an on-device GenAI model and Neural Processing Unit (NPU) to process all that data, making it searchable.

• When you perform a “Recall” action, the feature presents a screenshot related to your question to jog your memory.

• “Team Copilot” is an AI agent that assists users in Microsoft Apps (e.g., Word, PowerPoint, and Excel) to streamline workflows and Copilot Studio for developers to deploy AI agents.

• “Phi-3-vision” is a multimodal model that combines language and vision capabilities. It boasts strong reasoning power while using limited computing, making it a cost-effective option for small ventures.

Why It’s Important:

• “Copilot+ Recall” assists you by considering your historical behaviors and data, sparking concerns over privacy risks and new attack vectors for bad actors.

• Microsoft claims that the “Recall” action has been designed to save data directly on the user’s device in an encrypted format. Nadella added, “This is my computer. This is my Recall. And it’s all being done locally.”

• The screenshots won’t leave your computer. However, a loophole could be that the AI model scanning those screenshots will make data vectors for Microsoft to train on.

🩺 PULSE CHECK

ANTHROPIC

🗺Anthropic Maps the Minds of LLMs

Anthropic published a research paper that successfully identified millions of human-interpretable concepts within Large Language Models (LLMs) neural networks.

Key Details:

• Neural networks (i.e., AI frameworks designed to mimic the human brain) are viewed as a “Black Box” that requires large amounts of data and computational resources to generate outputs.

• It’s not clear why LLMs generate a specific output instead of another. Opening the “Black Box” doesn’t help either because LLMs “think” before generating responses, which appear as a series of numbered lists called “neural activations” in the data without a clear meaning.

• Anthropic researchers used a technique called “dictionary learning” to isolate patterns that corresponded to topics.

• By analyzing the patterns, Anthropic researchers were able to alter Claude 3 Sonnet’s outputs, possibly leading to more controllable AI models.

• The team mapped AI safety-related content, including deception, power-seeking, and privacy.

Why It’s Important:

• If we can understand how LLMs choose to generate outputs, we can reduce the number of harmful, biased, untruthful, and dangerous responses.

• By opening the “Black Box” and identifying meaningful concepts within AI’s workings, Anthropic can build safer and more reliable AI systems.

AI TIPS

🎓Google’s Free AI Certification Courses

Google just released a series of beginner-friendly AI certification courses for FREE, with no prerequisites or fees required.

Here Are 5 Beginner-Friendly AI Certification Courses You Can’t Miss:

1. Introduction to Generative AI

2. Introduction to Large Language Models (LLMs)

3. Introduction to Responsible AI

4. Introduction to Image Generation

5. Introduction to Machine Learning (ML) and AI on Google Cloud

🛠TRENDING TOOLS

📊Arcwise supercharges your Google Sheet with AI analytics.

📓Granola is the AI notepad for people in back-to-back meetings.

🌮Findr generates AI-powered answers using your workplace apps.

🔋Unify combines all AI models for faster, cheaper, and better responses.

💬Langotalk deploys chat-based AI tools to transform your language learning.

🔮Browse our always Up-To-Date AI Tools Database.

💰VENTURE CAPITAL UPDATES

• Bolster raises a $14M Series B to elevate its multi-channel threat protection.

• PatronusAI secures a $17M Series A to tackle hallucinations and copyright violations.

• Intently raises a $3M Seed Fund to launch an AI-powered networking tool for startup founders.

💼WHO’S HIRING?

• Grammarly (Remote): Applied Research Scientist Intern, Summer 2024

• ImpelAI (Syracuse, NY): Marketing Intern, Fall 2024

• Databricks (Mountain View, CA): IT Software Engineering Intern, Fall 2024/Spring 2025

• Bolster (Santa Clara, CA): Software Engineer, Backend Systems, New Grad

• NVIDIA (Santa Clara, CA): Technical Program Manager Intern, Business Intelligence, Fall 2024

🤖PROMPT OF THE DAY

DIGITAL PRESENCE

🚀Growth Hacker

Pretend you’re a growth hacker. Suggest initiatives to drive growth and boost my online presence for [Product/Service] with [Budget], [Current Growth Metrics], and [Goals].

Product/Service = [Insert Here]

Budget = [Insert Here]

Current Growth Metrics = [Insert Here]

Goals = [Insert Here]

📒FINAL NOTE

If you found this useful, follow us on Twitter or provide honest feedback below. It helps us improve our content.

How was today’s newsletter?

• ⭐️⭐️⭐️⭐️⭐️Nailed it!

• ⭐️⭐️⭐️Average

• ⭐️Epic fail.

❤️TAIP Review of the Day

REFER & EARN

🎉Your Friends Learn, You Earn!

You currently have 0 referrals, only 1 away from receiving ⚙️Ultimate Prompt Engineering Guide.

Refer 9 friends to enter 🎰May’s $200 Gift Card Giveaway.

Copy and paste this link to others: https://theaipulse.beehiiv.com/subscribe?ref=PLACEHOLDER