The Most INSANE AI News This Week! 🤯

Key Concepts

• OpenAI's image generation in ChatGPT
• Google's Gemini 2.5 Pro
• DeepSeek V3
• Microsoft 365 Copilot updates (Researcher, Analyst, Agent Flows)
• Model Context Protocols (MCPs)
• AI Agents for web automation
• Vibe coding
• AI-driven SEO

OpenAI's Image Generation in ChatGPT

• Main Point: OpenAI released a new image generation feature within ChatGPT that allows users to create Studio Ghibli-style images and perform advanced image editing with text prompts.
• Details:
  • Users can upload an image and request a style transfer (e.g., Studio Ghibli).
  • The AI understands images deeply, enabling specific edits, object additions, background changes, and lighting adjustments.
  • Example: Transforming a YouTube thumbnail into a Ghibli-style image.
  • The feature is available on ChatGPT Plus ($20/month) and Pro ($200/month) plans.
  • The free version was delayed due to server overload from high usage.
• Significance: This tool democratizes high-quality image creation, making it accessible to users without professional graphic design skills. It acts as a creative partner, understanding composition, aesthetics, and design principles.

Google's Gemini 2.5 Pro

• Main Point: Google released Gemini 2.5 Pro, a powerful AI model that outperforms other top models in benchmarks and is free to use via AI Studio.
• Details:
  • Gemini 2.5 Pro surpasses Claude, GPT-4, and other leading AI models in various benchmarks.
  • It features a million-token context window, allowing it to process approximately 750,000 words (several books) at once.
  • It excels in coding tasks, generating cleaner and more functional code than models like GPT-4.
  • Example: Creating 3D racing simulators, interactive physics demos, and dinosaur-themed running games with single prompts.
  • Example: A Rubik's Cube simulation where colors are perfectly tracked during rotations, a task other AI models struggled with.
  • Supports Vibe coding, enabling conversational code generation.
  • Cost: Free in AI Studio, compared to GPT-4.01 Pro which costs $600 per million tokens.
• Significance: Gemini 2.5 Pro offers unparalleled performance and accessibility, making it a game-changer for developers and researchers. Its large context window and coding capabilities open up new possibilities for AI applications.

DeepSeek V3

• Main Point: DeepSeek, a Chinese AI company, released DeepSeek V3, an open-source model performing at the level of GPT-4.5 and Claude 3.7.
• Details:
  • Released under the MIT license, making it completely open-source with no API costs or usage limits.
  • Can run at 20 tokens per second on an M3 Ultra Mac.
  • Users are building SAS tools, games, and applications with it.
  • Example: Building an SEO keyword research tool with a user-friendly interface in minutes.
  • Available on the DeepSeek website, LM Arena, and OpenRouter.
  • Can be used with Visual Studio Code through extensions like CLaient.
• Significance: DeepSeek V3 democratizes access to high-end AI, enabling developers to build AI applications without significant financial investment.

Microsoft 365 Copilot Updates

• Main Point: Microsoft introduced new features in Microsoft 365 Copilot, including Researcher, Analyst, and Agent Flows, enhancing data analysis and automation capabilities.
• Details:
  • Researcher and Analyst: Use OpenAI's 03 mini reasoning model for advanced data analysis, processing complex datasets, visualizing information, and creating reports with Chain of Thought reasoning.
  • Agent Flows: Allow users to create custom agents for specific business needs, enabling mini-agents to run on proprietary business data.
• Significance: These updates empower businesses to leverage AI for data-driven decision-making and automation of complex tasks.

Other Notable AI Developments

• Model Context Protocols (MCPs): OpenAI is adopting MCPs, a standard introduced by Anthropic for connecting LLMs to tools on the internet, creating a standardized layer between AI and software APIs.
• Google Meet and Maps Updates: Google added features to Google Meet for capturing follow-up action items and suggesting next steps. Google Maps now allows saving screenshot locations for travel planning.
• Claude 3.7 Sonnet: Anthropic is reportedly working on giving Claude 3.7 Sonnet a 500,000-token context window.
• Perplexity: Introducing new search filters for images, video, travel, and shopping.
• Luma AI: Showcased Magic Doodles, allowing animation of hand-drawn images.
• Pabs: Rolled out a flashback feature where a photo version enters a video frame.
• Earth AI: Using algorithms to find critical minerals in overlooked locations.
• Waymo: Expanding self-driving vehicle services to Washington D.C. next year.
• Boston Dynamics: Demonstrated advanced robotics capabilities, including robots running, crawling, and performing barrel rolls.

AI Agents for Web Automation

• Main Point: AI agents can automate web tasks, such as data scraping, form filling, content creation, and report building, without requiring coding knowledge.
• Example: Using the browser AI agent in a tool called RODE inside Visual Studio Code.
• Process:
  1. Install the RODE extension in Visual Studio Code.
  2. Connect it to an API like Claude 3.7 Sonnet.
  3. Instruct the AI agent to perform specific tasks, such as posting a tweet or conducting competitor analysis.
• Example: Using the AI agent to perform keyword research with Ahrefs, identifying low-competition keywords for a dentist and generating a comprehensive report for approximately $0.20 worth of API tokens, compared to $250-$500 charged by an SEO agency.
• Significance: AI agents streamline web-based workflows, saving time and resources for businesses and individuals.

Conclusion

The AI landscape is rapidly evolving, with significant advancements in image generation, language models, and automation tools. OpenAI's image generation in ChatGPT, Google's Gemini 2.5 Pro, and DeepSeek V3 are democratizing access to powerful AI capabilities. Microsoft's Copilot updates and the emergence of AI agents for web automation are transforming business workflows. These developments are making AI more accessible, practical, and integrated into everyday life, empowering creators, developers, and businesses to achieve more with less effort. The pace of innovation is accelerating, making it crucial to stay informed and adapt to these transformative technologies.