Welcome, AI Enthusiasts!

Gmail’s new AI-driven search surfaces emails based on relevance, not just keywords. Claude gains web search capabilities, delivering real-time information with citations. Google’s budget-friendly Pixel 9A loses key AI features due to hardware constraints. And in an innovative twist, a high schooler built a Minecraft-based benchmark to evaluate AI creativity and reasoning.

In today’s Generative AI Newsletter:

• Gmail’s AI Upgrade – Smarter email search prioritizes relevance.

• Claude Gets Web Search – AI chatbots now retrieve real-time info.

• Pixel 9A’s Gemini Limitations – Budget AI comes with missing features.

• Minecraft as an AI Benchmark – A creative way to test AI’s intelligence.

✨ Cut Development Time in Half with AI-Powered Low-Code Tools

What if you could bring your innovative ideas to life—without writing endless lines of code? Does faster builds, smarter solutions, and less manual coding sound too good to be true?

With AI-powered low-code tools, you can make it a reality.

Join mindit.io’s Expert-Led Webinar

"AI-Powered Low-Code: Faster Development, Smarter Innovation"

Hosted by Laurentiu Amitroaie (CINO) and Radu Bostina (UI Designer & Marketing Graphic Designer), this 45-minute session will show you how AI is transforming low-code platforms.

You’ll learn how to:

 ✅ Accelerate development cycles with AI automation

 ✅ Enable rapid prototyping to test ideas quickly

 ✅ Overcome challenges like customization and scalability

📅 March 27 @ 2:00 PM CET

Ready to supercharge your development process?

👉 Sign up now and unlock the power of AI-powered low-code tools.

🔍 Gmail’s AI Upgrade: Smarter, Faster Email Searches

For years, Gmail’s search has worked like a basic filter—sorting emails chronologically based on matching keywords. But Google’s new AI-powered update rethinks search entirely, shifting from a static keyword lookup to a dynamic ranking system that understands what matters most to you.

What’s New?

1.  Relevance Over Recency – Instead of treating all keyword matches equally, AI ranks emails based on importance, considering:

🔹How often you interact with a contact

🔹Which emails you click on the most

🔹The recency of the conversation

2. Adaptive Learning – Over time, Gmail’s AI learns from your search behavior, refining results so that the emails you’re actually looking for surface faster.

3. Dual Search Views – You can still switch back to the old chronological method, but the AI-powered “Most Relevant” mode aims to reduce the number of searches needed.

Before, if you searched for “invoice,” Gmail would dump every email with ‘invoice’ in it, ordered by date. Now, AI prioritizes the invoices you actually open and interact with, bringing them to the top.

The inbox is no longer just a passive storage space; it’s learning how you work and adjusting to you.

🌐 Claude Can Now Search the Web

Anthropic’s AI chatbot, Claude, just gained a game-changing capability—web search. Unlike traditional chatbots that rely purely on pre-trained data, Claude’s AI now works like a real-time research assistant, dynamically retrieving information from the web to inform its answers. It doesn’t just return search results but also processes them, summarizes key insights, and delivers responses in a conversational, contextual format

 How Web Search Works:

• Smarter Query Handling – Claude decides when web search is necessary, ensuring only relevant queries trigger it.

• Fact-Checked Responses – Web-based answers include direct citations, making it easy to verify sources.

• Conversational Search – Unlike search engines that return a list of links, Claude processes and summarizes results into coherent, AI-generated insights.

• Use Cases – This update benefits market analysts, researchers, sales teams, and even everyday users looking for real-time information.

Chatbots are moving beyond static knowledge toward real-time learning, making them more like true digital assistants. With Claude, OpenAI’s ChatGPT, and Google’s Gemini all racing toward smarter, more responsive AI, the future of AI assistants is becoming clear.

 📱 Pixel 9A’s Gemini AI is Missing Key Features

Google’s budget-friendly Pixel 9A comes with a stripped-down version of Gemini AI, lacking some of the smart features found in pricier Pixel 9 models. The Pixel 9A runs a smaller on-device AI model, Gemini Nano 1.0 XXS, which only activates when needed instead of running in the background. This makes a noticeable impact on AI-powered features:

What’s Different?

• No Pixel Screenshots – Unlike higher-end models, the Pixel 9A can’t use AI to make screenshots searchable.

• No Call Notes – AI-generated summaries of phone calls won’t be available.

• Recorder App Summaries Still Work – Since these rely on transcriptions, they remain functional.

• Memory Matters – The Pixel 9A has 8GB of RAM, while the Pixel 9 has 12GB and the Pixel 9 Pro models have 16GB.

Google made a similar move last year when only the Pixel 8 Pro initially got Gemini AI, but later rolled it out to the regular Pixel 8 after user backlash. Their approach shows that on-device AI still depends heavily on hardware, even as AI models get more efficient. The question is whether users will accept feature limitations on budget devices or push for broader AI access like they did with the Pixel 8.

🏗 This High Schooler Is Using Minecraft to Test AI Models

Traditional AI benchmarks often fail to capture real-world performance, so 12th-grader Adi Singh built MC-Bench, a website where AI models compete to generate Minecraft builds. Users vote on the best creations without knowing which AI made them, making evaluation more intuitive than code-based benchmarks.

How It Works:

• AI-Generated Minecraft Builds – Models write code to create structures from prompts like “Frosty the Snowman.”

• Blind Voting – Users vote before seeing which model made each build, reducing bias.

• Supported by AI Giants – OpenAI, Anthropic, Google, and Alibaba subsidize MC-Bench’s use of their models but aren’t directly involved.

• Beyond Simple Builds? – Singh sees potential for testing long-term planning and reasoning in games, where AI can be evaluated more naturally than through static tests.

AI can ace standardized tests but struggle with simple reasoning—MC-Bench provides a more practical test. As AI models become more goal-driven, interactive benchmarks like this could offer a clearer picture of their real-world intelligence. If a high schooler can push AI evaluation forward, should the industry be rethinking its approach?

🚀 Boost your business with us—advertise where 10M+ AI leaders engage

🌟 Sign up for the first AI Hub in the world.

📲 Our Socials