Today, much of the research we do happens online. Whether for work, school, or something personal – like finding the best new headphones – we do it online. AI has been increasingly shaping that process, but before we talk about its role, it helps to look at how we got here. 

Not too long ago, videos and memes circulated online making fun of older adults for asking Alexa how she was doing before making their request or typing into the Google search bar that they would like its help. This seemed like an obvious waste of time to us at the time, a way to muddy your search results. The accepted way to conduct research was to type in a couple of words into your search engine: “Halloween costumes” or “boutique hotels in Puerto Rico”, or even “intercultural communication”.  

You had to think of all the possible key words and all the combinations of those words together. You would look through the search results, page after page.

Since then, things have shifted. Search Engine Optimization, PageRank, and KnowledgeGraph took the world by storm. And the world beyond the first three pages of a Google search result stopped existing for the average search or user.

It seemed like distilling your concepts to a few keywords was the essence of the research process.

However, with the rise of generative AI, we now look at those same grandmothers who were being polite with Alexa, and we have to acknowledge: she knew what she was doing.

You see, even though the evidence is preliminary and highly dependent on model, language and task, pleasantries and politeness with a chatbot improve its performance such as: response accuracy and less misunderstanding (Piperski, 2024)). In fact, talking to chatbots disrespectfully can lead to less responsiveness from it (Quan & Chen, 2024). Therefore, your grandmother was right. When you conduct research with AI, you want to not only be nice to your chatbot, but also provide it with all the information we once deemed unnecessary in a Google search.

You want to move far beyond keywords and give your chatbot a paragraph or two that includes not just pleasantries, but also context, goals, and even the role you want it to take on. This is what we call prompting. Prompting is one of the main ways you interact with generative AI.

AI has been changing the job market including and not limited to research. AI automation has led to a 23.4% reduction in traditional middle-skill jobs, generated a 31.7% increase in new employment categories, with proactive reskilling programs achieving a 64% higher retention rate (Kanagarla, 2024). This makes prompting an increasingly valuable skill, especially in roles that involve research, analysis, or decision-making. The good news is that these are skills most people can strengthen.

When I talk to my clients, I always try to find what skills they already have that are transferable to research with AI and prompting specifically. I have worked with K-12 literature teachers who teach students how to build background knowledge through answering questions like: Who? What? Where? Why? When and How? This type of context is exactly what a good prompt provides for an AI chatbot. Similarly, project managers are already good at defining goals, constraints, deliverables. Prompting is what allows us to interact with AI and help us augment our research process. It is the first building block to shifting your research process to utilize AI. General guidelines to improve your AI prompting game involve keeping a clear goal in mind for what you are trying to achieve and a clear task you are looking for the chatbot to start with for that goal. Provide your chatbot with the research context and do not forget to select the appropriate thinking power in your bot for the depth of your research. Clear goal, clear task, enough context.

As the research process adapts to the AI tools which have been inserting themselves into workflows across industry and academic positions – research continues to guide us through the best prompting practices. Research papers and thought pieces are available and most relevant to each field. Consider looking into Park & Choo (2024) if you are in the education field, if you are in the medical industry – Varia et al. (2025) and Velasquez-Henao et al. (2023) for engineering.

Finally, reach out to me for support in training your team to prompt effectively while also integrating AI into your business or university in ways that are practical, ethical, and built for real-world use. If you are ready to figure out how your current skills can be extended into good prompting, you can find me at Denison Edge with workshops this summer and at this website: sheylafinkelshteyn.com