From the course: AWS Certified AI Practitioner (AIF-C01) Cert Prep

Unlock this course with a free trial

Join today to access over 24,900 courses taught by industry experts.

Prompt engineering risks and limitations

Prompt engineering risks and limitations - Amazon Web Services (AWS) Tutorial

From the course: AWS Certified AI Practitioner (AIF-C01) Cert Prep

Start my 1-month free trial Buy for my team

Prompt engineering risks and limitations

“

- With prompt engineering, while the constant tinkering with those prompts can absolutely lead to higher accuracy, better overall performance of the model, there are some risks and limitations. The first of these is exposure, and this is where sensitive information could be inadvertently exposed because of the prompt engineering that's being performed. And so a poorly crafted prompt, it might encourage the AI to generate confidential data, which could violate privacy laws. But a more specific example, if the AI itself is exposing details about its own algorithm because of an improperly designed prompt, that can be a big deal. Next we have poisoning, and this is where the intentional manipulation of the model's training data or inputs can induce harmful behavior. And so, as a generic action, you just have an adversary submitting a large number of biased or misleading prompts during training to corrupt the behavior. But more specifically, it's been shown that attackers can subtly modify…

Contents