The MacKellar Memo | May 2023
Working with AI: Come with me if you want to live!
As the days grow longer and flowers begin to bloom, we dust off our bikes, find the paddleboard paddles and the, just as we’re ready to go enjoy the amazing outdoors here in Colorado, it [INSERT RANDOM SPRING WEATHER THING, LIKE HAIL, SNOW, SLEET, ETC.] So, while you’re completing the fun Colorado madlibs of what weird weather ruined your plans, you think what else should I do in spring? Well, you’re stuck inside due to [RANDOM SPRING WEATHER] so you might as well do some spring cleaning, right? But this wouldn’t be a marketing-related article if we didn’t have a pivot from the fun intro into our topic, so we’ll go as subtle as Ross moving a couch in an NYC walkup — PIVOT! Just as we declutter our homes and refresh our spaces, it's also an excellent opportunity to fine-tune our marketing communications practices. With the increasing use of AI-powered tools in our industry that have been flying around fast and furious in our AI Exchange, let's dive into the world of prompt engineering and explore how it can transform the way we approach our work given its role as one of the backbones of AI.
What is Prompt Engineering?
Prompt engineering is a technique used in natural language processing and artificial intelligence, particularly with language models like GPT-4. It focuses on crafting effective input prompts that guide AI models to produce desired and relevant responses or outputs. The key to successful prompt engineering lies in clarity, context, constraints, and examples. By refining and iterating prompts, AI developers and users can improve the quality of responses generated by language models, making them more useful and effective in various applications.
A prompt might look like the following examples:
Poor prompt: "Write about AI."
Better prompt: "Write a 500-word introductory article about the applications of artificial intelligence in the marketing industry."
The better prompt is more specific and provides clear context and constraints, guiding the AI to produce a more relevant and targeted response.
Prompt Engineering + You
Understanding and applying prompt engineering effectively is crucial for leveraging AI-powered tools in marketing communications. As AI systems become increasingly sophisticated, learning how to communicate with them efficiently and precisely is vital for optimizing their performance and maximizing their potential. The right prompts can help guide AI tools to generate more accurate, relevant, and useful outputs, streamlining marketing tasks and ultimately leading to better results.
With this in mind, here’s a few tips for writing effective prompts that can enhance our use of AI in our everyday marketing work:
Be specific and clear: Use precise language and avoid ambiguity. Clearly define your expectations, which will help the AI understand your intent and generate more accurate responses.
Provide context: Include necessary background information or context to help the AI generate a relevant response. For example, if you want the AI to write about a new product launch, provide details about the product and target audience.
Set constraints: Define specific constraints, such as the format of the response, tone, or style, to ensure the output aligns with your expectations. For instance, specify the desired word count (or token on OpenAI), tone (formal or informal), and target audience (industry professionals or general public).
Include examples: Providing examples within the prompt can help guide the AI to produce a response that is more in line with the desired outcome. For example, if you want the AI to create catchy email subject lines, provide a few examples of successful subject lines you've used in the past.
Iterate and refine: It's essential to test and refine your prompts to optimize the AI's performance. Analyze the AI-generated responses and adjust your prompts accordingly to achieve better results.
String together multiple prompts: Break down complex tasks into a series of simpler prompts, allowing the AI to tackle each part separately before combining the results. This approach can lead to more coherent and accurate outputs. For instance, if you want the AI to draft a marketing campaign proposal, start by asking it to outline the campaign goals, followed by separate prompts for target audience analysis, marketing channels, and budget allocation.
While today is just the beginning of the “new” AI in our daily work, prompt engineering plays (and likely will continue to play — just don’t make it a career path, apparently) a crucial role in harnessing the power of AI for marketing communications. By embracing the spirit of freshening up our approach to things each spring and applying these tips, we can refine our approach to prompt engineering, leading to more effective and successful outcomes in our AI-powered marketing endeavors.