AI responds to your emotions

It might sound strange but AI actually does perform better when you say something emotional IN ALL CAPS.

Learn

Emotion Prompting

Making emotional appeals actually works in prompts because LLMs have learned from the internet to respond more diligently when people are emotional...More

Experience

Mike Taylor

Built a 50-person growth agency.
Logo
Logo
Logo
💪 Useful 0
😓 Difficult 0
🎉 Fun 0
😴 Boring 0
🚨 Errors 0
😕 Confusing 0
🤓 Interesting 0
Premium subscription required.
Python experience recommended.
1. Scenario
UPSERT HEADQUARTERS - MARKETING STRATEGY MEETING
Alright, team, gather 'round! We've got a new strategy to discuss that's going to take our marketing efforts to the next level. We're going to dive into the fascinating world of emotion prompting and how it can boost the performance of our language models. By incorporating psychologically motivating language and emotional cues, we can expect a small but significant increase in accuracy, around 5 to 10%. So get ready to experiment with different techniques and test, test, test! We'll gather all the data we need and analyze the results to see just how effective this approach can be. Alright, folks, let's get started and unleash the power of emotion in our AI!
Sally Valentine
at Upsert

Alright, team, gather 'round! We've got a new strategy to discuss that's going to take our marketing efforts to the next level. We're going to dive into the fascinating world of emotion prompting and how it can boost the performance of our language models. By incorporating psychologically motivating language and emotional cues, we can expect a small but significant increase in accuracy, around 5 to 10%. So get ready to experiment with different techniques and test, test, test! We'll gather all the data we need and analyze the results to see just how effective this approach can be. Alright, folks, let's get started and unleash the power of emotion in our AI!

This course is a work of fiction. Unless otherwise indicated, all the names, characters, businesses, data, places, events and incidents in this course are either the product of the author's imagination or used in a fictitious manner. Any resemblance to actual persons, living or dead, or actual events is purely coincidental.

2. Brief

Emotion Prompting: Enhancing Language Models with Psychological Stimuli

Emotion prompting is an intriguing technique that I find both enjoyable and effective. It is a simple method that yields noticeable improvements in accuracy, typically resulting in a 5% to 10% increase. The implementation process is straightforward, making it highly accessible. In essence, emotion prompting involves incorporating emotionally stimulating elements from psychology into large language models (LLMs) to enhance their performance.

The underlying principle behind the success of emotion prompting lies in the recognition that people respond differently when they encounter emotional cues. This phenomenon has been observed in online interactions, where individuals tend to pay more attention and respond more diligently when emotions are involved. For instance, when someone emphasizes the importance of a message by using capital letters or even threatens harm if a response is not provided, individuals tend to react more promptly.

While such strategies were more prevalent in the past, as modern LLMs have become better at following instructions, they still hold value and are worth exploring. As an example, a developer once tested the effectiveness of emotion prompting by requesting a language model to output the full code with the plea that they lacked fingers. Surprisingly, the model complied, demonstrating the power of emotion prompting in action.

To illustrate the concept further, let's consider a simple example. Suppose we have a language model with a "get completion" function and we ask it to provide a detailed explanation of photosynthesis spanning 2000 words. Typically, without emotion prompting, most LLMs would fall short, generating only around 1000 words. However, by incorporating an emotion prompt, we observe a significant improvement. In this case, the word count reaches 1069, showcasing the impact of emotion prompting on performance. It's important to note that these results may vary due to the non-deterministic nature of the models.

To conduct a comprehensive evaluation, it is recommended to run multiple tests and average the results. This can be achieved by setting up asynchronous runs, allowing for simultaneous testing of both standard prompts and emotion prompts. By doing so, we can gauge the average word count and determine the effectiveness of emotion prompting compared to traditional prompts.

During a test with 30 runs, it became evident that emotion prompting consistently outperformed standard prompts, resulting in a 12% increase in word count. This improvement demonstrates the potency of incorporating human emotion and motivation into the prompt.

In conclusion, emotion prompting is a simple yet powerful technique for enhancing language models. By leveraging

3. Tutorial

  Emotion prompting is one of my favorite techniques because it's quite fun. And easy and it just works. Quite often, it's only a small percentage increase typically in accuracy. Maybe around 10%, five to 10%, but it's very easy to implement. And what we're talking about when we talk about emotion, prompting is specifically. This paper a large language model has come stand and can be enhanced by emotionally stimuli.

EmotionPrompting.ipynb
Download
4. Exercises
5. Certificate

Share This Course