A clear and concise prompt entered into any generative artificial intelligence (generative AI) software is the difference between an accurate and effective response or an error-riddled, unusable mess.
‘Prompt crafting’ or ‘prompt engineering’ is becoming a crucial science as more accounting and finance professionals leverage the capabilities of generative AI. Generative AI has myriad uses and that’s why the technology became so popular so quickly within organisations and financial institutions. Management accountants can programme the software and be freed to do other, high-level strategic tasks. And you can strengthen your proficiency in strategic problem-solving and digital technology by earning the Chartered Global Management Accountant® (CGMA®) designation.
But it all goes back to the prompt.
Here are five tips for fine tuning your generative AI prompts and tapping into the seemingly infinite possibilities of generative AI technologies.
Define expectations with detail and specificity
Generative AI tools like ChatGPT can work their magic with a simple prompt like ‘analyse the following communication history’. From the results, you can decide next steps.
But detail and specificity can be used to your advantage. Unlike traditional AI, generative AI is adaptable and can create new content; a longer, detailed prompt unlocks those capabilities.
Your search could then become, ‘Search the following memo for all instances of the word “supply chain”, place the full paragraph of each instance into a bulleted list and add relevant web references from large organisations and academic sources’.
Ask for variations
Don’t settle for the first response; instead, request several variations. This gives you more ideas to consider.
Generative AI is often used to write emails or marketing communications. You can request three or four versions and choose which one, or which components, work best for your purposes.
Alternatively, asking for variations increases the effectiveness for research. You can ask for more results than three or four depending on your research needs. For example, ‘What are the top 10 highest tax rates in UK history’?
Request citations and explanations
Earlier iterations of generative AI tools were prone to ‘hallucinations’ — adding false facts into their answers. You can catch the lie by asking the tool to explain itself.
As you’re crafting your prompt, you can include script that instructs the tool to ‘provide direct citations to sources’. Then you can check the list of sources to ensure accuracy. Or you could check the logic by asking the tool to explain its chain of thinking.
Beyond fact-checking, requesting citations and explanations are great educational tools. Depending on generative AI’s coding, you could ask questions to seek clarity on an advanced programme, like an Excel macro.
Learn like AI
Prompt engineering is a science, and it takes time to refine. You may not get your desired response the first time around, but simple tweaks could yield better results.
Add a word. Change a phrase. Don’t abandon your prompt just yet — any minor change will alter the output. Keep massaging the prompt to see what you come up with.
As you’re crafting your prompts, ask your team members for their best writing tips and tricks.
Test out advanced capabilities
Once your prompt-engineering skills feel strong, push the limits of generative AI.
Instead of asking questions, you can request the tool to pose questions to you. For example, you could input ‘Ask me questions that will help me understand the breadth of my organisation’s carbon footprint’, or, trying something more personal, ‘Ask me questions to define my leadership style’.
These questions positively challenge your critical thinking skills and your conventional ways of thinking.
Depending on the generative AI tool, you could use the bot’s conversational skills to practice talking to a client, colleague or manager — and then have it analyse your performance.
Generative AI streamlines processes. With that in mind, you can experiment with different advanced techniques to see how they can be applied to accounting.
Generative AI is expanding what management accountants can do and how they do their jobs. Generative AI is a great performance enhancer to pre-existing technologies.
As more financial organisations implement generative AI and other advanced technologies, there’s increased demand for professionals who understand generative AI models and additional digital technologies. Management accountants can harness the power of digital tools and apply them directly to their day-to-day tasks.
Ready to level up your career? Kick-start your digital development, solidify your organisation’s digital future and see how the CGMA designation is essential to navigating this tech-driven world.
