By providing more detail, you can enhance the AI's ability to generate accurate responses. To demonstrate, begin by inputting "Write me a story about a bookstore" into ChatGPT and observe the output. Next, input the aforementioned prompt and witness the disparity in the results.
You can see how providing more detail allows the AI to generate better responses. To demonstrate this, start by inputting "Write me a story about a bookstore" into ChatGPT and observe the result. Then, use the above prompt and compare the difference.
3. Direct the AI to assume specific identities or professions
One of ChatGPT's impressive features is its ability to write from the perspective of a particular person or profession. In the past, I showcased how ChatGPT could mimic a pirate's or Shakespeare's style of writing. However, the AI can also adopt the voice of a teacher, a marketing executive, a journalist, or any other role you desire.
Here's an example. Let's have ChatGPT describe the Amazon Echo smart home device from the viewpoints of a product manager, an adult child caring for an elderly parent, and a journalist. These are the three prompts:
Describe the Amazon Echo Alexa device from the perspective of its product manager.
Describe the Amazon Echo Alexa device from the perspective of an adult child caring for an elderly parent.
Describe the Amazon Echo Alexa device from the perspective of a journalist.
You can input these prompts into ChatGPT to view its complete responses. However, I'll share a few sentences to illustrate how ChatGPT takes on different roles to offer distinct perspectives.
When writing as a product manager, the AI responded with this sentence: I can confidently assert that this is one of the most innovative and revolutionary products in the smart home industry.
From the perspective of a caregiver, the AI's response included: The device's capability to set reminders and alarms can be particularly beneficial for elderly individuals who may struggle to remember taking their medication or attending appointments.
As a journalist, the AI's response contained: From a journalistic standpoint, the Echo has garnered attention due to privacy concerns surrounding the collection and storage of user data.
As you can see, different perspectives enable the AI to offer diverse insights within its responses. I will delve further into this concept by demonstrating how to engage the AI in a thought experiment. Let's explore the considerations involved in developing a device like Alexa:
Imagine it's 2012. Siri has been available for the iPhone for about a year, but there hasn't been anything similar to the Alexa smart home device. Picture an Amazon board meeting where the Echo, a smart assistant based on Alexa, has just been proposed.
Detail the arguments that the board members at that meeting might have discussed in deciding whether or not to invest in the development of such a device, considering both pros and cons.
You may even involve engineering design experts and product champions for a more comprehensive perspective.
Here's a helpful tip: making slight modifications to your prompts can significantly influence ChatGPT's response. For instance, when I adjusted the phrase "Detail the arguments, pro and con, that..." to "Provide the pro and con arguments in the form of a dialogue that...," ChatGPT rewrote its answer, transitioning from a list of enumerated pros and cons to an actual conversation among participants.
Additional tips for crafting prompts
- Feel free to rephrase the question since ChatGPT often alters its response with each subsequent query.
- Make minor adjustments to your prompts to guide ChatGPT towards providing better answers.
- ChatGPT retains its knowledge of previous conversations as long as you keep the current page open. However, if you close the page, ChatGPT loses its contextual understanding. It's worth noting that sometimes ChatGPT loses track of the conversation anyway, so you may need to start over occasionally.
- Starting a new page initiates a fresh discussion with ChatGPT.
- Long answers exceeding approximately 500 words can sometimes lead to breakdowns. Therefore, be sure to specify the desired length of the response.
- If the AI misinterprets your query, you can correct and clarify your prompt based on its previous answer. This helps it understand what it missed and proceed accordingly.
- Shape your questions differently if ChatGPT is hesitant to answer as you intended. Utilize personas to elicit responses it might otherwise be reluctant to provide.
- If you require sources and citations, instruct ChatGPT to support or substantiate its answers.
- Continue experimenting with different prompts and approaches.
- Occasionally, ChatGPT may fall short. While persistence is key, be open to utilizing alternative tools if necessary. ChatGPT is not flawless... yet.
Which prompts yield the best results with ChatGPT?
One of the remarkable aspects of ChatGPT is its versatility—it can respond to almost any question. Nonetheless, keep in mind that ChatGPT specializes in generating written answers. If you seek a list of websites, it's best to consult Google. For computational tasks, turning to Wolfram Alpha is more appropriate. To get the most out of ChatGPT, provide open-ended prompts that encourage creativity (yes, it turns out creativity isn't only reserved for humans). Feel free to share personal experiences or emotions, and remember that the AI's knowledge is up to date as of 2021.
What should I do if ChatGPT refuses to answer or I'm dissatisfied with its response?
ChatGPT has certain guardrails in place. For instance, it tends to avoid answering political questions. While it's possible to coax out a response, the value it provides may be limited. However, don't hesitate to rephrase or approach the question from different perspectives.
You can ask as many questions as you like (though it once told me I had asked too many and suggested I come back later). Yes, it basically said, "Go away, kid, you're bothering me." To be fair, after hours of questioning, I probably was indeed bothering it. This tool can be as addictive, if not more so, than watching puppy videos on YouTube.