Keeping up with the rapid pace of technological advancements can feel overwhelming, but learning how to use Artificial Intelligence (AI) can simplify and enhance many aspects of your working day. To harness AI’s full potential, it is crucial to know how to interact with its tools effectively. This article will guide you through a step-by-step process on the best practices for prompting AI tools.

Creating Effective Prompts

1. Understand Your Goal:

Identify the purpose of your prompt. Are you seeking information, generating creative content, or problem solving? Clear goals lead to more precise prompts and ultimately better results.

• A way to gain general information is to write a sentence structured like the following, “what advantages have emerged due to remote working since it became prevalent during the COVID-19 pandemic?”
• If you desire creative content, you could ask the AI to “write a short story about an accountant who gains the superpower of becoming a mathematical genius every time he works remotely.”
• If you require help with solving a problem, you could ask a question posed like the following sentence, “when I’m working from home, how can I improve my time management skills for better productivity?”

2. Provide Clear and Specific Context:

Giving background information and precise context helps the AI generate more relevant responses. For example, “Explain the societal impact of COVID-19 on office workers in the United Kingdom between the years 2019-2022” is more effective than simply asking “what was the impact of COVID on office workers”. The more precise your prompt, the better the AI can understand and respond accurately.

3. Set Parameters:

Specify the format, length, and/or style of the response. Using the ‘COVID-19’ prompt mentioned above, I used co-pilot to generate a response. The AI then replied with numerous lengthy paragraphs, and although it gave me the information I needed, the amount of information was excessive, and it wasn’t written in the format I had imagined. Therefore, I continued my conversation by sending the following request, “using that information, please write me a 150 word summary in British English, in a formal tone” as shown below.

4. Use Structured Queries:

Break down complex requests into simpler, structured parts using proper grammar and punctuation. AI is different to search engines such as Google or Bing – AI is made to be used interactively, as if you’re talking to a friend or a colleague, so it’s important to write in complete sentences to ensure the AI can accurately interpret your prompt.

After your initial request, you can then ask the AI to modify or adjust its output piece by piece to guide it until you reach your desired outcome. Continuing with our example, I then prompted the AI to “remove the section about hybrid work models, and expand on the point regarding mental health”, which generated the following result:

5. Role-Playing Personas:

Asking the AI to adopt a specific persona can yield more tailored and insightful responses. For example, the prompt “you are a staff member of a company who is writing a formal complaint via email to the Directors, no longer than 100 words, due to them pressuring you to attend the office when you have COVID”, generated the format that I needed, instead of the long paragraphs that it made in our original prompt. This is all due to the way that the interaction is structured. Afterward, you can then use the aforementioned points to tailor the results to best match your needs.

6. Check, Check, Check:

When using AI, it’s response could contain false or misleading information that is presented as fact, and has the ability to integrate false information with reality, making it impossible to discern between fact or fiction. There is also the growing concern that as AI advances rapidly, the amount of human-created data available for training these models is becoming insufficient, meaning that much of the content produced by AI is also being consumed from other AI data. This increases the likelihood of the data being incorrect or being retrieved using illegitimate sources. Therefore, not only is it vital to ask AI to only use valid and reliable sources when collecting information from the internet, but to also fact check everything by performing your own research, so you are able to cross reference and be comfortable with the validity of the information.

Tips for Beginners

Start Simple: Begin with straightforward prompts and gradually add complexity as you become more comfortable.

Learn from Feedback: Pay attention to the AI’s response and use them to improve your future prompts. If the initial response isn’t what you expected, experiment! You can adjust the wording, or phrase your text differently, sometimes a slight change in wording can make a big difference.

Stay Patient: It might take more than a few tries to get the perfect response, so don’t be discouraged!

Remember, AI can be a valuable tool that can help enhance your ideas, but it’s important to understand its limitations and not overestimate its capabilities. We should aim to collaborate with AI by refining the content it generates, experiment to gain the best results, whilst always fact checking throughout the entire process.

As you continue to refine your prompts, you’ll find that these tools can become invaluable assets in various tasks, from content creation to problem-solving. Remember, effective communication with AI is a skill that improves with practice, so keep exploring and don’t be discouraged.

Stay tuned for the second half of this series, titled ‘Avoiding Common Mistakes and Ensuring Accuracy in AI Prompting’, coming soon!

This article will guide you through a step-by-step process on the best practices for prompting AI tools.