• AI Fire
  • Posts
  • 💔 Microsoft Two-Timing OpenAI...

💔 Microsoft Two-Timing OpenAI...

PLUS: Claude's Intern-Killing Upgrade

In partnership with

ai-fire-banner

Read time: 5 minutes

Google built an AI that outperformed human scientists even at writing real scientific methods. It beat top experts in genomics, COVID modeling, neuroscience, and more.

LEARNING PARTNER AIRCAMPUS

Create your own AI Agents👨‍🏭👩‍🏭

While most people juggle tabs and chase tasks, a few are handing it all off - to AI agents that don’t assist, they act. Imagine:

  • Your email replies are done. 📬

  • Instagram posted. 📸

  • Calendar managed. 🗓️

  • 40+ apps humming in the background. ⚙️🤖

  • All without lifting a finger. 🧘‍♂️

On Thursday, 11th September at 10 AM EST, we're pulling back the curtain.

It’s an exclusive Masterclass showing you how to architect AI agents that actually run your work - across 40+ tools - so you can focus on the parts only you can do.
This is your invite.
🎟️ Save Your Seat – Before It’s Gone

AI INSIGHTS

googles-ai-just-out-researched-the-researchers

What would happen if you gave an AI the job of writing scientific methods instead of essays or code snippets? Google Research just tested that idea and the results are wild. Here’s how it works:

  • You feed it a scorable task (problem description, data, and evaluation metric)

  • The system generates ideas → writes code → tests thousands of variants using a tree search loop to improve scores, keep the best ones

  • End result: expert-level, reproducible solutions in hours instead of months

The benchmarks it crushed:

  • Genomics → 40 new RNA-seq integration methods, top one 14% better than the best expert tool

  • COVID-19 → Outperformed the CDC’s gold-standard CovidHub Ensemble with 14 new forecasting models

  • Neuroscience → Predicted activity of 70K zebrafish neurons better than all baselines

  • Mathematics → Solved 17/19 hard integrals where standard methods failed

  • Time-series → Auto-generated a forecasting library across 28 datasets

It’s essentially an AI-powered research engine that can explore 1,000x more possibilities than humans alone.

Why It Matters: Most coding AIs just try to be “syntactically correct.” Google’s is idea-generating + performance-maximizing. And it never gets tired, never stops exploring, and can push entire fields forward at warp speed.

PRESENTED BY ROKU

Kickstart your holiday campaigns

CTV should be central to any growth marketer’s Q4 strategy. And with Roku Ads Manager, launching high-performing holiday campaigns is simple and effective.

With our intuitive interface, you can set up A/B tests to dial in the most effective messages and offers, then drive direct on-screen purchases via the remote with shoppable Action Ads that integrate with your Shopify store for a seamless checkout experience.

Don’t wait to get started. Streaming on Roku picks up sharply in early October. By launching your campaign now, you can capture early shopping demand and be top of mind as the seasonal spirit kicks in.

Get a $500 ad credit when you spend your first $500 today with code: ROKUADS500. Terms apply.

TODAY IN AI

AI HIGHLIGHTS

📝 NotebookLM users can now auto-generate custom reports, full blogs, flashcards, quizzes, and video summaries in 80+ languages. See the new features in action here.

🎬 OpenAI is backing Critterz, a full AI-generated animated film to hit Cannes 2026. It already made its debut in a 2023 animated film with a short video to watch here.

🗂️ Claude can create & edit PDFs, slide decks, reports with visual charts, full write-ups from raw data or spreadsheets for you now in chat. Here’re some ways to use them.

🎮 Google with Stanford just launched AI Quests, a fun game series for middle schoolers. It’s “edutainment” for AI. Kids role-play as Google researchers. Try it for free

😆 If you call ChatGPT ‘dumb’, you'll get roasted, not by the bot, but by other humans like this. One user even asked if that makes them a bad person & got a witty answer.

🎙️ Inception Point plans to produce 5,000 AI-generated podcasts per week, with 3,000 new episodes weekly at just $1 per one. Critics are screaming “soulless content”!

💰 AI Daily Fundraising: Cognition AI raised $400 million, reaching a $10.2B valuation in 18 months. Their tool, Devin, grew ARR from $1M to $73M. Clients include Goldman Sachs and Cisco.

AI SOURCES FROM AI FIRE

ai-fire-academy

AI CHEAT SHEET

NEW EMPOWERED AI TOOLS

  1. 🤖 Empromptu builds full-stack, AI-native applications in minutes

  2. 🚀 Revenuee launchs high-converting pages & builds your brand

  3. 📊 Chart2Data turns any chart image into clean, usable data instantly

  4. ✍️ ProdPocket crafts high-quality content to boost search rankings

AI QUICK HITS

  1. 🔥 Microsoft Azure OpenAI offers new conversational AI shopping

  2. 📢 Microsoft will use Claude Sonnet 4, in shift from OpenAI GPT-5

  3. 🤖 Apple is brewing 'World Knowledge Answers', a ChatGPT rival

  4. 🔍 AI researchers often spotlight well-cited works & certain names

  5. 🍎 Apple new iPhone 17 devices don’t have an AI-powered Siri yet

AI CHART

alibaba-wants-to-be-the-new-standard-for-speech-recognition

Alibaba just unveiled Qwen3-ASR-Flash, a speech recognition model built on Qwen3-Omni with some clever new tricks!

What Qwen3-ASR-Flash Can Do:

  • Outperforms rivals in Chinese, English, and 9 more languages

  • Can transcribe rap and songs even with background music

  • You can feed it keywords, docs, or nonsense text to bias results without preprocessing

  • Works under vehicles, crowd chatter, code-switching, or lecture audio

  • Detects language automatically and ignores background noise/silence

Most ASR models are good, but rigid. They can’t flex to your context without major engineering. Qwen3-ASR is basically prompt engineering for speech recognition.

→ Want your transcripts to always get your brand name right? Feed it the docs once, and you’re set.

OpenAI’s Whisper dominates open-source ASR. Qwen3-ASR is Alibaba’s enterprise-ready challenger, with singing + accent handling as its differentiator.

Alibaba’s trying to own how the world talks to AI. And if they nail this, voice might be the arena where Qwen finally leapfrogs Western models.

We read your emails, comments, and poll replies daily

How would you rate today’s newsletter?

Your feedback helps us create the best newsletter possible

Login or Subscribe to participate in polls.

Hit reply and say Hello – we'd love to hear from you!
Like what you're reading? Forward it to friends, and they can sign up here.

Cheers,
The AI Fire Team

Reply

or to participate.