Why is IT support AI so bad at routing tickets?
AI ticketing tools frequently misroute requests and fail at basic tasks like password resets, making administrators unwilling to grant them permissions. These security risks and poor accuracy mean teams cannot rely on automation without creating more work for themselves.
What people tried
Every workaround mentioned in the threads below. We haven’t tested any of them — and nobody here is claiming they worked.
- 1Using traditional automation for password resets and account access instead of AI
- 2Restricting AI usage to suggesting resolutions or searching knowledge bases while keeping human validation
- 3Using AI purely for initial information collection and finding similarities in past tickets without execute permissions
In their words
Unedited, grouped by where they were said, most upvoted first within each place, each linked to the thread it came from.
“The amount of tickets we are getting asking if we can make other people's jobs easier through ai is getting ridiculous.”source ↗
“So why are you asking me to get AI to scan your documents and input them in a sheet?”source ↗
“Not only no, but hell no lol”source ↗
“Ah yes, let's have the AI reset passwords. 🥸”source ↗
“most stories that I heard about AI ticketing systems were very bad.”source ↗
“Yes. They did a terrible job with it. It rarely routes a ticket accurately”source ↗
“Giving it Execute permissions would be a no-go for us.”source ↗
Where this came up
People with this problem also raised
- 2How to manage too many AI automation ideas
- 4How do solo sysadmins handle endless ad-hoc software and system requests?
- 20Why does fixing AI mistakes take longer than doing it myself?
- 22Why is project management just doing admin and updating trackers?
- 2Should service desks track ticket effort?
- 3How to choose an AI customer support agent for real use cases