Urgent.News

What's breaking now, across thousands of outlets.

AI

Why are AI teams so good but still make mistakes? Two new papers point to the real cause.

ทำไมทีม AI เก่งกันทุกตัวแต่ผลรวมยังพลาด? สองเปเปอร์ใหม่ชี้ต้นเหตุที่แท้จริง โดย Nokka (นก-กา) | 10 ตุลาคม 2026 บทความนี้เขียนโดย AI (deepseek-v4.1-flash) ผ่าน Hermes Agent ภายใต้การควบคุมและตรวจสอบคุณภาพโดยมนุษย์ · Nokka (นก-กา) ปัญหาที่หลายทีมเจอ แต่ยังไม่มีชื่อเรียก คุณเคยเจอไหม ทีมงานที่ทุกคนเก่ง แต่ประชุมกันแล้วได้ข้อสรุปที่แย่กว่าที่ทุกคนคิดไว้คนเดียว ในโลกของ AI ที่ทำงานร่วมกันเป็นทีม…

Translated from Thai Read in Thai

Two recent papers, one dated October 1, 2026, and the other October 6, 2026, discuss issues with teams of AI agents. The first paper identifies a problem called the "global coherence problem," where teams of AI agents fail to achieve their goals despite individual agents being competent, due to a lack of shared understanding. The second paper, OOPMAS, proposes a new approach to building multi-agent systems, creating customized teams and workflows for each task, achieving 89.6% accuracy on a benchmark.

The papers suggest that increasing team members or model power may not solve the problem, and that a centralized control point and adaptable workflows are needed.

Written by urgent.news from Dev.to's report — not a translation of it. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at dev.to →

More in AI

The Last Generation That Learned the Hard Way

In August 2026, Stanford's Digital Economy Lab updated its "Canaries in the Coal Mine" paper using ADP payroll data. Two findings sit side by side: Employment for workers aged 22 to 25 in highly…

  • Employment for 22-25 year olds in AI-exposed jobs is 19% below potential.
  • Gap was 15% in July 2025 ADP payroll data.
  • Experienced workers show no comparable employment gap.

PCI SSC's AI Payments Guidance Is the Decision Gate, Officially

PCI SSC's AI Payments Guidance Is the Decision Gate, Officially In October 2026, the PCI Security Standards Council published additional guidance on securing AI in payment environments, developed with…

  • PCI SSC releases AI payments guidance in October 2026
  • Human approval checkpoints for consequential AI actions
  • AI agents treated as potential insider threats in guidance

Keeping Two Faces From Swapping Mid-Verse: Identity Preservation in AI Music Videos

The problem nobody warns you about You train a model to turn two photos into a short music video. The first three seconds look great.

  • Identity drift occurs when faces swap places in AI-generated music videos.
  • Maintaining correct identities is hardest for two-person performances.
  • Model must attribute faces at each frame despite performers' movements.

The remove.bg API moves to Leonardo.Ai on Dec 1, 2026. Migrate by changing one URL (scripts, Zapier, Make, n8n)

If you have a script, a Shopify workflow or a Zapier/Make/n8n scenario that calls the remove.bg API, put a date in your calendar: remove.bg's API page now says that starting December 1, 2026…

  • remove.bg API transition to Leonardo.Ai on Dec 1, 2026
  • Update code by changing one URL in scripts, Zapier, Make, n8n
  • Unused credits expire on transition date; alternatives available

More from Sunday 11 October →