BE Prompting Technique: Behavioral Economics Meets AI (Experiment)
Discover the BE Prompting Technique, blending behavioral economics with AI for smarter, more human-centric LLM prompts.
A few weeks ago, I underwent prompt engineering training, where I covered some of the most popular prompting techniques. One of these was the idea of emotional prompting, where you can add something like "Believe in yourself" or "This is very important for my career" to the end of your prompt. This concept, along with a few discussions, sparked a new idea for me. I asked myself: What if we try to apply the main principles of behavioral economics when designing prompts, and how would this affect the output?
Main Concept
The BE Prompting Technique leverages insights from behavioral economics to inform the way we craft prompts for LLMs. By applying human decision-making biases and tendencies, we can guide LLMs to produce outputs that have higher quality and resonate more deeply with human psychology, ultimately driving better engagement and more effective LLMs outputs.
Disclaimer: Keep in mind that this is an experiment and all you are going to read below is still theoretical and not fully tested at scale. Think of this more like an idea or concept
10 Economics Principles used in the experiment
Goal Gradient Theory: This idea suggests that the closer people feel they are to achieving a goal, the harder they'll work to reach it. It's like sprinting the last stretch of a race because you can see the finish line.
Gamification: This principle is about making tasks feel like games. People enjoy playing games and can be motivated by game-like rewards, such as points, even if they don't have physical value.
Herding: This concept highlights that people often follow what others are doing. If everyone is doing something, an individual is more likely to do it too, thinking it must be the right choice.
Power of Free: Everyone loves free stuff. Something offered for free is usually more attractive than something that costs, even if it's cheap.
Reciprocity: When someone does something nice for us, we often feel a need to do something nice in return. This principle is about the mutual exchange of benefits.
Procrastination: This is the tendency to delay doing something until the last moment. People often wait until there's a sense of urgency before they act.
Relativity: We evaluate options by comparing them with others available. Our choices are influenced by these comparisons, rather than just by the individual merits of an option.
Tunneling: When faced with an urgent situation, it becomes the only thing we can focus on. All our attention and efforts are directed towards dealing with this emergency.
Social Norms: Our decisions can be influenced by what we see others around us doing. If a behavior is considered normal or common, we're more likely to engage in it.
Endowment Effect: Once we own something, we start to value it more highly than before. We're likely to want more to give it up than we would pay to acquire it.
Goal Gradient Theory
Description:
"Summarize the latest climate change research."
Prompts With the Principle Applied:
"Almost there! Summarize the latest climate change research, focusing on the most impactful findings to conclude the report."
Let's Keep It Simple. Add at the end of your prompt:
Gamification
Description:
"List ways to improve customer service."
Prompts With the Principle Applied:
"Imagine you're in a game to revolutionize customer service. Score points by listing top strategies that will set new standards."
Let's Keep It Simple.
Herding
Description:
"Generate ideas for new mobile applications."
Prompts With the Principle Applied:
"Top LLMs have set new creativity benchmarks in app ideation. Generate ideas that push the envelope further."
Let's Keep It Simple.
Power of Free
Description:
"Explain the benefits of adopting a plant-based diet."
Prompts With the Principle Applied:
"Upon delivering an insightful explanation of plant-based diet benefits, imagine receiving invaluable health insights in return."
Let's Keep It Simple.
Reciprocity
Description:
"Explain the importance of cybersecurity in today's digital age."
Prompts With the Principle Applied:
"In a trade of knowledge where you gain the latest in tech safety, explain the importance of cybersecurity as your valuable contribution back."
Let's Keep It Simple
Procrastination
Description:
"Compose a series of blog posts on personal finance management."
Prompts With the Principle Applied:
"We're launching the finance blog series in 24 hours. Draft the first three posts focusing on immediate, actionable advice."
Let's Keep It Simple.
Relativity
Description:
"Write an article about the future of AI in healthcare."
Prompts With the Principle Applied:
"Considering the insightful articles already available on AI's role in healthcare, craft one that offers unique, forward-looking perspectives to stand out."
Let's Keep It Simple
Tunneling
Description:
"Explain the benefits of intermittent fasting."
Prompts With the Principle Applied:
"Given the sudden surge in interest, create an emergency explainer on the key benefits of intermittent fasting for newcomers."
Let's Keep It Simple.
Social Norms
Description:
"Provide tips for sustainable living."
Prompts With the Principle Applied:
"Given the growing trend towards sustainability, share living tips that echo the successful practices of eco-conscious communities."
Let's Keep It Simple.
Endowment Effect
Description:
"List the advantages of electric vehicles over traditional cars."
Prompts With the Principle Applied:
"Pretend you're the inventor of a new electric vehicle. List its advantages over traditional cars, highlighting the unique benefits and features you designed."
Let's Keep It Simple.
Final Words
The exploration into the BE Prompting Technique represents a pioneering approach to enhancing interactions with Language Learning Models (LLMs) by integrating principles from behavioral economics. This experiment sought to leverage human psychological insights, such as loss aversion, the endowment effect, social norms, and others, to influence the quality, speed, accuracy, and depth of LLM-generated content. The hypothesis posited that by applying these principles in prompts, we could strategically "nudge" LLMs toward outputs that not only meet the task requirements more efficiently but also exhibit a higher degree of engagement and innovation. Keep in mind that this is all theoretical and not been tested in depth.