AI idea generation has so many benefits in market research. It’s easy to use, it’s cheap, and it actually works pretty well… as long as you leverage the right tactics. We’ve been playing with artificial intelligence tools for a while now and want to share 1 big tip for making the most out of them and getting better results.
“What’s the tip?” you might ask. It’s to give AI tools more prompts.
Spot the bot! Some parts of this blog post were written using ChatGPT, a new AI tool created by OpenAI, which was trained to interact with humans via chat. The tool can support a dialogue, answer questions in detail, create content and even write code. See if you can find parts of this blog post that ChatGPT wrote for us.
The basics: what are prompts?
If you ever use artificial intelligence in market research, prompts will be your best way to communicate with AI-powered tools.
Prompts in AI are instructions or cues that provide the necessary context and guidance for AI to perform a specific task or give you a response. Prompts can be many things: natural language instructions, structured data inputs, or pre-defined rules and parameters.
For example, a prompt can be "provide a summary of the latest news stories related to Apple." In this case, the prompt gives AI the guidance to identify and summarize relevant news stories.
The basics: why do you need prompts?
You might be thinking, “Hey, AI is already smart enough. Why should I do the work myself and give it examples?”. Well, it’s true that AI is smart. After all, it uses machine learning to store and educate itself from huge amounts of data.
But just like you needed some direction when you were learning math in first grade, AI needs direction too. This direction is provided through prompts. Here’s why you need them.
1) AI will have a clear understanding of what is expected of it
Prompts can help AI tools to see what kind of ideas you need exactly and what format you’re looking for (e.g. long sentences vs short phrases).
2) AI can focus on a specific task or objective
This way, it’ll give you the most relevant and useful information. This is especially important when there is a large amount of data. Prompts can help AI filter out the most relevant info.
3) AI can give responses that are relevant for specific audiences
Let’s say you’re looking for flavor names for a new energy drink for teenagers. Typical flavor names for this market usually sound edgy and cool, like “Fizz Frenzy”, “Sonic Boom” or “Bubble Gum Mania” (ChatGPT gave us these examples!). So, of you want AI to create similar names for your specific market, you’d want to give it examples / prompts that follow a similar structure and are tailored to the needs and interests of this specific group of consumers.
Prompts in practice: Sneaker ideas
To show you how useful a good prompt can be in AI, we decided to ask our AI tool for some innovative sneaker ideas.
The prompt was: “Give me 10 new sneaker ideas.”
This is what the tool generated:
1. Light-up soles
2. Magnetic lacing
3. Self-drying fabric
4. Waterproof materials
5. Recyclable materials
6. Temperature control features
7. Motion-sensing technology
8. Shock-absorbing foam
9. Flexible sole design
10. Anti-microbial linings
Not bad, not bad. But do any of these ideas make you excited? Maybe not so much. They kind of feel standard, like any sneaker brand would probably have something similar.
But what happens when we give AI a bit more context in the prompt?
“Give me 10 new sneaker ideas. Example: Back to the Future self-lacing shoes.”
1. Motion-sensing shoes that adjust to the terrain for a more comfortable walking experience.
2. Eco-friendly vegan sneaker materials.
3. Color-changing shoes with a variety of designs.
4. High-tech running shoes with GPS tracking capabilities.
5. 3D-printed shoes that can be customized to the wearer’s foot.
6. A fashion sneaker with a built-in sound system.
7. Breathable mesh sneakers designed to keep feet cool and dry.
8. A sneaker with a built-in air filtration system.
9. A sneaker with a built-in solar panel for charging devices.
10. A sneaker with a built-in LED display for notifications.
Okay, now we’re getting something. While some of these shoes kinda sound whacky and probably not feasible (like #9, with a solar panel to charge devices), they still are innovative. And it would have taken a lot of time for a person to come up with these ideas. AI completed the task in a few seconds.
Artificial intelligence needs human intelligence
AI definitely takes the load off market researchers’ shoulders by offering innovative ideas in a matter of seconds. But there’s still a long way for market research to fully rely on AI technology. These tools need human intelligence to direct them (i.e. prompt) so that the results are more accurate and efficient.
