I broke ChatGPT.

Well, not for everyone…but it completely shut down on me. Zero output after 6+ hours of working with it and waiting.

This experience perfectly illustrates why we can’t rely too heavily on AI for answers or miracles. I’m seeing more discussion about Gen AI’s limitations around original thinking, creativity, and fundamental learning (check out Krys Boyd‘s Think podcast or recent posts by former Kalypso: A Rockwell Automation Business colleagues Steve Hurst and Vesy Ivanova below).

𝗛𝗲𝗿𝗲’𝘀 𝘄𝗵𝗮𝘁 𝗵𝗮𝗽𝗽𝗲𝗻𝗲𝗱:
While writing a recent white paper with Noel Sobelman, I used AI to identify examples supporting our core message. The problem? When we fact-checked, almost all the sources were broken or potentially fabricated—classic hallucinations.

When I asked ChatGPT to verify sources and find valid replacements, it crashed. After 6 hours of “amazing output coming soon” messages and apologies, I gave up and manually researched everything case-by-case using different platforms.

The lesson? Always 𝗰𝗵𝗲𝗰𝗸 𝘆𝗼𝘂𝗿 𝘄𝗼𝗿𝗸  and 𝗻𝗲𝘃𝗲𝗿 𝘀𝘁𝗼𝗽 𝘁𝗵𝗶𝗻𝗸𝗶𝗻𝗴 𝗳𝗼𝗿 𝘆𝗼𝘂𝗿𝘀𝗲𝗹𝗳.

Have you experienced similar AI breakdowns?

𝗥𝗲𝗳𝗲𝗿𝗲𝗻𝗰𝗲𝗱 𝗽𝗼𝘀𝘁𝘀:
·      Noel and my Article on AI for CPG Innovation: https://lnkd.in/g7NFRBjP
·      Vesy’s Post: https://lnkd.in/gGmGJKKM
·      Steve’s Post: https://lnkd.in/gdeSpgnH
·      Think Episode: https://lnkd.in/gb_dxDFh