Urgent.News

What's breaking now, across thousands of outlets.

AI

I just did something my AI agents couldn't

It is funny how things work out. You can let AI tools try to fix a code problem for days, but nothing works until you finally step in and do it yourself. I was working on my ESP-IDF project and wanted to organize it better. My goal was to move different pieces of code into a folder named components . Try #1: Gemini 3.6 Flash I don't pay for AI, so I use the free web versions. I used a tool I…

I embarked on a quest to organize my ESP-IDF project, aiming to relocate various code segments into a folder titled "components." Armed with the free version of AI tool ReptClip, I copied my project files and then instructed the AI to determine which files needed relocating and what new code should be inserted. Regrettably, the AI's suggestions resulted in a cascade of red error lines across my code editor, with the entire file eventually inundated by errors.

Undeterred, I sought assistance from the Pro version of the AI. Despite employing the same sequence of steps, the outcome remained identical - a deluge of errors. This experience was repeated multiple times, with me growing increasingly frustrated and even cursing the AI for its ineptitude.

In a bid to break free from the cycle of failure, I ventured forth to explore other AI models. Gemini Flash Extended, Gemini Pro Extended, DeepSeek v4 Flash, and DeepSeek v4 Pro were all subjected to the same treatment, yet none managed to resolve the issue. In total, I tried an indeterminate number of iterations, ultimately spending two entire days pursuing a solution that proved elusive.

Undaunted, I turned to GitHub Copilot, entrusting it with the task. To my dismay, it too took approximately thirty minutes to complete the job, only to leave behind a trail of error squiggles. I then took a brief respite, returning to the project with renewed vigor. This time, I chose to bypass the AI's recommendations and tackle the issue directly.

After careful deliberation, I recognized the root of the problem: the #include paths. Armed with this insight, I communicated my predicament to the AI, requesting assistance with the REQUIRES or PRIV_REQUIRES settings. While the AI provided some guidance, it still misidentified the exact headers required, leading me to conduct additional research online.

With the correct REQUIRES line in place, my project finally compiled successfully. However, it became increasingly apparent that the AI had consistently provided erroneous advice throughout the ordeal.

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

[Insight] Why Nvidia Is Bringing Rival AI Chips Into Its Ecosystem

U.S. AI chip startup d-Matrix said on Sept. 10 that its next-generation Raptor XPU will connect directly to Nvidia MGX rack-scale infrastructure through NVLink Fusion.

  • Nvidia integrates rival AI chips, d-Matrix's Raptor XPU, into MGX infrastructure.
  • Raptor's memory-centric architecture reduces data movement bottlenecks in AI inference.
  • Partnership enables Nvidia's GPUs and Raptor to work together on AI inference stages.

More from Saturday 12 September →