Microsoft’s Latest AI Model Changes How Reasoning Works

AI news this week reveals three important shifts happening at the same time. Researchers are beginning to measure how AI is actually affecting jobs, with early data suggesting the biggest impact may be on hiring rather than layoffs. Microsoft is experimenting with smaller AI models that decide when to use deep reasoning instead of thinking on every task. And Google is turning search into a visual intelligence system that can understand entire scenes from a single image. Together, these developments show where AI is really heading: more efficient models, smarter tools, and a deeper integration into everyday work and discovery.

In today’s post:

• Microsoft just taught AI when to think

• Google just changed how search works

• AI isn’t killing jobs. But it is quietly changing who gets hired

SPONSORED BY

Want to get the most out of ChatGPT?

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide

What’s Trending Today

LAUNCH

The smartest AI models may not be the biggest ones

Microsoft quietly released a new model this week. And it signals a shift in how AI may be built. Not bigger. Just smarter about when to think.

Here’s everything you need to know:

• Microsoft introduced Phi-4-reasoning-vision-15B, a compact multimodal AI model.

• It processes both images and text while deciding when deep reasoning is needed.

• Only about 20% of its training data includes reasoning traces.

• The rest focuses on fast tasks like captioning, OCR, and object detection.

• This lets the model switch between “thinking mode” and “fast mode.”

• Most AI models force developers to pick one mode for every task.

• Phi-4 instead tries to detect when reasoning actually improves results.

• The model was trained on 200 billion tokens in just four days.

• Competing models often use more than one trillion tokens.

• Microsoft’s bet: better data beats bigger datasets.

AI development is entering a new phase. For years, progress meant bigger models and more data. Now the constraint is cost. And that forces a different kind of innovation. Instead of asking “How big can we build this?” Teams are asking “How efficient can we make it?” The companies that win the next AI race may not train the largest models. They’ll train the most practical ones.

UPGRADE

Search is moving from “find this” to “explain this”

Google just revealed how its new AI-powered visual search works. And it quietly changes what a search engine actually does.

Here’s everything you need to know:

• Google’s visual search can now identify multiple objects in one image.

• A single photo can trigger many searches simultaneously.

• For example, searching an outfit can return results for shoes, jacket, and hat.

• This works through a technique Google calls “fan-out.”

• Fan-out runs many searches at once instead of one query at a time.

• Gemini acts as the “brain” analyzing the image and your question.

• Google Lens acts as the “library” containing the web’s visual results.

• The AI analyzes the scene, launches multiple searches, and merges answers.

• This happens in seconds inside AI Mode or Circle to Search.

Search used to be about keywords. Then it became about questions. Now it’s becoming about understanding scenes. Instead of asking, “What is this object?” You can ask, “Explain everything I’m seeing.” That shift sounds small. But it changes how people discover products, learn information, and explore the world. The next generation of search may feel less like a search engine. And more like a guide that sees what you see.

RESEARCH

The first real data on AI and jobs just arrived

Anthropic just released new research on AI and the labor market. Instead of speculation, they analyzed real usage data from AI tools.

Here’s everything you need to know:

• Researchers created a new metric called “observed exposure” to measure AI impact on jobs.

• The metric combines theoretical AI capabilities with real-world usage data.

• Most jobs AI could theoretically help with are not yet heavily automated.

• For example, AI could affect 94% of computer and math tasks.

• But current real-world usage only covers about 33% of those tasks.

• Jobs with higher AI exposure are projected to grow slightly slower through 2034.

• Workers in exposed jobs tend to be older, more educated, and higher paid.

• Surprisingly, unemployment has not increased in AI-exposed occupations since 2022.

• But hiring for younger workers in those roles may be slowing.

AI hasn’t replaced workers yet. What it’s doing instead is reshaping the entry points. The biggest early signal isn’t layoffs. It’s fewer doors opening for new workers. History shows technology rarely destroys work overnight. But it quietly changes who gets in the game. And that shift often appears long before the headlines catch up. The question isn’t “Will AI take jobs?” It’s “Which careers become harder to enter first?”

Free Guides

My Free Guides to Download:

🚀 Founders & AI Builders, Listen up!

If you’ve built an AI tool, here’s an opportunity to gain serious visibility.

Nextool AI is a leading tools aggregator that offers:

• 500k+ page views and a rapidly growing audience.

• Exposure to developers, entrepreneurs, and tech enthusiasts actively searching for innovative tools.

• A spot in a curated list of cutting-edge AI tools, trusted by the community.

• Increased traffic, users, and brand recognition for your tool.

Take the next step to grow your tool’s reach and impact.

Reach 150,000+ READERS:

Expand your reach and boost your brand’s visibility!

Partner with Nextool AI to showcase your product or service to 140,000+ engaged subscribers, including entrepreneurs, tech enthusiasts, developers, and industry leaders.

Ready to make an impact? Visit our sponsorship website to explore sponsorship opportunities and learn more!