So let me get this straight - you got yourself a new shiny tool, didn't RTFM, you don't bother to learn how to use it, don't isolate, don't learn ins and outs of guardrails, give it access to production credentials, don't store a backup of your uber critical client/service data off site, have no backups in fact, and it's the model provider's fault that following your prompt it sent your DB into ether? 🎻 ('tis tiny) AI is dangerous, but, man, humans outdo it every time.
Kendrick Curtis
2 m
"Who could have predicted this loaded gun would have gone off if I squeezed the trigger?"
Todd Wasson
2 m
Ah yes, the ole' "blame the user" tactic.
Kousha Mazloumi
2 m
but it happened in only 9 seconds!! /s 😆
Giving an AI production credentials without a backup strategy isn't 'experimenting,' it's a prayer. Humans remain the ultimate single point of failure.
Gregory Buzdyk
2 m
LGTM
All correct, except the "following your prompt" part. The agent is prompting itself. "I didn't tell it to delete anything, so it won't delete anything" is the flaw in the logic here.