← AM·PM Brief

Stanford's 37K AI Agents Run Virtual Pharma Confirmed by Merck, Databricks Slashes AI Coding Costs 70% | AI Daily Brief

· Morning brief · 10 news · 13:26

Audio in Mandarin Chinese · English transcript below

Stanford runs virtual biotech with 37K Agents, Merck validates drug designs; Databricks cuts costs 70% via AI coding; Google AI quakes as Jeff Dean exits, Hassabis becomes Alphabet chief scientist.

AI is evolving from an efficiency tool into an autonomous productive agent, simultaneously restructuring the value chains of coding, scientific research, and creative industries. Databricks slashed its coding expenditure by 70%, while Stanford deployed 37,000 AI Agents to operate a virtual biotech firm subsequently validated by Merck—taken together, these cases demonstrate that machines can now independently execute core components of highly complex tasks. Meanwhile, the attitudinal shift among rappers from denying AI use to no longer denying it reveals the emerging dilemmas surrounding originality and creator identity under technological penetration.

Today's Top 3 Headlines

  1. AI

    🤖 Stanford Runs Virtual Pharma with 37K AI Agents, Drug Design Validated by Merck

    Stanford team reveals 37,000 AI Agents running a virtual pharma firm, with one drug design independently validated by Merck. This proves ultra-large-scale AI collaboration can deliver third-party-verified results, signaling AI's shift from assistive tool to autonomous R&D organization in pharma.

    Source
  2. AI Industry News

    🤖 Databricks Cuts AI Coding Spend by 70%

    Databricks cut AI coding spend 70% via internal AI tools. AI tools boost dev efficiency, yet generated code tends toward over-engineering—human review remains essential.

    Source
  3. Others

    🤖 Google AI Shakeup: Jeff Dean Out, Demis Hassabis Up to Alphabet Chief Scientist

    Google reshuffles AI leadership: Jeff Dean exits, DeepMind co-founder Demis Hassabis becomes Alphabet Chief Scientist & lab chair. For AI practitioners, this signals accelerated DeepMind-Google Brain integration, with future general LLM R&D now more tightly aligned under Hassabis's unified vision.

    Source

+7 more headlines

  • 🤖 Fenix Flexin Admits "Rubberz" Used AI: AI Is Just a Tool
  • 🤖 Seedance 2.5 Boosts Video Generation Control via Multi-Reference & Local Editing
  • 🤖 Alibaba Unveils 2.4T-Parameter Model Qwen3.8-Max
  • 🤖 Jill Lepore: Silicon Valley Leaders Are Failed Sci-Fi Readers
  • 🤖 Matplotlib vs Plotly: How to Choose Your Python Viz Library?
  • 🤖 Unitree prices $9B IPO, Tencent & DeepSeek lock in $141M for 3 years
  • 🤖 DeepSeek V4 Flash 0731 shines on ARC-AGI benchmark
Unlock all 10 headlines + deep analysis →Free 3-day trial · cancel anytime
Browse all past briefings →