Urgent.News

What's breaking now, across thousands of outlets.

AI

While AI Is Thinking

You wake up. Your brain isn’t quite awake yet, but social media already seems to be having a very productive morning. “Here’s what you can do with AI.” “If you don’t know this trick, you’re missing out.” “This made me ten times more productive.” There they are again, all lined up for the day. You think, wow, that sounds useful. So you try it. And, usually, it doesn’t work quite that well. Maybe…

You awaken and your mind is disoriented, yet social media appears to be already hard at work in the morning. "Here's what you can do with AI." "If you don't know this trick, you're missing out." "This boosted my productivity tenfold." These messages appear repeatedly, beckoning you to try them. And, typically, they don't live up to the hype.

Perhaps I'm using them incorrectly. Perhaps they truly help the person who shared them, but I suspect the simpler explanation is that no one truly knows yet. We are still searching for the right approach.

There isn't a clear way to utilize AI. We aren't aware of how to effectively request its assistance. We aren't certain about what to provide to it and what should remain our responsibility. It feels more like everyone is searching for the right way, rather than someone having discovered it. Consequently, it becomes somewhat tiresome when someone claims to have found "the right answer" each morning.

It's understandable that big claims capture attention more effectively. However, the individuals making these claims often come across as overly optimistic. "Everything has changed!" "There's no turning back!" "If you don't learn this now, you'll fall behind!" And I can't help but wonder: Does the world truly need to change so frequently before the first coffee?

What I genuinely desire to understand is something less grandiose. For instance: You ask AI to perform a task. And the screen displays: "Thinking..." For ten seconds. Maybe thirty. What should a human do during this time? Shake out your shoulders. Gaze out the window. Sip some water. Close your eyes. Contemplate something completely unrelated. Instead of teaching me how to accelerate AI, I would prefer someone to teach me how to cope while AI is pondering. I would read that article immediately.

Perhaps the Empty Time is the Point. If AI provides us with a bit more time, perhaps we don't need to fill that time with something else. Maybe enhancing convenience doesn't mean being able to accomplish more tasks. Perhaps it could mean something smaller and better. Maybe convenience is having a little extra time in which nothing needs to be done.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

How We Built a YouTube SEO Pipeline with AI Agents

A multi-agent system that automates video research, scripting, rendering, upload scheduling, and analytics — fully self-hosted. The Problem Running a YouTube channel is a full-time job.

  • Multi-agent system automates YouTube SEO pipeline
  • Three AI agents coordinate tasks via message bus
  • Offline TTS and FFmpeg used for video rendering

Upgrading Your Embedding Model Doesn't Break RAG Loudly — It Breaks It Quietly

The problem A team I was helping upgraded their embedding model to cut cost — swapped an older general-purpose embedding model for a newer, cheaper one.

  • Upgrading embedding model to cheaper one reduces costs without schema changes or errors.
  • Retrieval systems return most similar vectors even if incorrect, no errors thrown.
  • Mixing embedding models in one index silently corrupts retrieval results over time.

Spotting Invisible LLM Agent Bugs with Agnost AI

Detecting Invisible Errors in LLM‑Powered Agents with Agnost AI Your practical guide to monitoring, debugging, and automating remediation in production pipelines Introduction When your autonomous…

  • Agnost AI detects invisible LLM agent errors in production
  • Monitors runtime state, tool calls, latency, token usage
  • Identifies issues over weeks, unlike traditional testing

Three AI Coding Agents, Three Ways to Break Them, and One Thing Detection Will Never Give You

Three AI Coding Agents, Three Ways to Break Them, and One Thing Detection Will Never Give You I've spent the last week reading through the technical writeups from Novee Security's Black Hat USA 2026…

  • Claude Code, Gemini CLI, and OpenAI Codex each have distinct vulnerabilities
  • Vulnerabilities stem from trust handoff failure in agent validation
  • Cryptographic evidence needed to prove trust handoff integrity

More from Wednesday 26 August →