Have you tried using AI to support your business, only to find yourself underwhelmed by the results? It can be too easy to assume that the AI will be able to extrapolate what we really mean based on vague instructions, but that just isn’t how it works.
In order to get quality results, you need to provide the AI tool with a quality request. This will require something called “prompt engineering.”
Let’s go over some strategies to help you make the most of the prompts you create.
For optimal effectiveness, a good prompt needs three things: context, specificity,and iteration.
Who are you asking the AI to be, and why?
Think of it like this: without any information to start from, the AI has no basis from which to craft an answer. It has no point of reference. Giving the AI a role to use and additional details provides it with the reference it needs, ultimately helping make your prompt more effective.
For instance, let’s say you wanted social media content to post to a local restaurant’s profiles to promote group reservations and events. Instead of simply prompting the AI to “produce social media content,” you should prompt it with something like the following:
“You are a marketing professional. Create content that speaks to New York business owners that highlights how [Restaurant Name] offers facilities and catering for holiday parties and other events. Make sure these posts are conversational and engaging, welcoming local businesses to enjoy the menu that [Restaurant Name] provides for groups.”
See the difference? More details give the AI more ingredients to work with, which ultimately benefits you.
Speaking of details, don’t skimp on giving the AI variables to consider. The more granular you can be with your prompts, the better, as it helps improve the quality of the results you generate. For instance:
If at first you don’t succeed, try, try again. Modern AI chats are designed to remember what you’ve entered and reference it as you continue working with your prompts. Therefore, repeating your prompts, reiterating and refining them, is the key to adjusting the outcome.
One tip is to treat your prompt like a conversation, following up on different points and giving the AI even more specific context to factor in. You might respond with feedback like, “This looks great, but I need it to be written for an audience of high schoolers,” or “Please expand that last bullet into a full paragraph.”
Repetition is the name of the game. Use the AI’s speed to power through multiple versions as quickly as possible, cherry-picking the best results from each run.
Have you ever met someone who is brazenly, confidently wrong? Someone who gets an idea in their head and treats it as gospel truth? AI is exactly the same, but in a few more insidious ways.
First, since AI is trained on human-supplied data, it inherits any biases present in that data. This can easily skew results, leading to bad info coming to you.
Second, AI is overly eager to please and will occasionally invent information to provide you with what you are looking for. This is known as hallucination, and means that you should always check an AI’s work.
It’s just crucial that you give it the right, specific context and iterate on the responses it produces. Like any technology, it all depends on what you put into it.
Suffolk Computer Consultants is here to help you manage all of your business’ technology. Give us a call at 631-905-9617 to learn more.
About the author
Suffolk Computer Consultants has been serving the Speonk area since 2013, providing IT Support such as technical helpdesk support, computer support, and consulting to small and medium-sized businesses.
Mobile? Grab this Article!
Tag Cloud
Newsletter Sign Up
Comments