
Prompt engineering, generative AI | Mihail Eric, founder of Confetti AI and Pametan Data Innovation
Infinite Curiosity Pod with Prateek Joshi
00:00
How Can I Trick an LLM?
Riley Goodside posted a few interesting screenshots on how it's possible to trick an LLM. The models are so good at obeying these prompts and kind of listening to them, he says. You can sometimes troll these models, tell it to do something completely unrelated to what it was trainedHow? By telling it to ignore its prior prompts,. Tell it to just echo something completely different back at you.
Transcript
Play full episode