I find myself going to ChatGPT several times a day to do a quick research on any topic, to bounce off my thoughts, or to create a first draft. On one side, while we keep thinking about enterprise-level use cases of ChatGPT, I have been wondering what it means for the skillset of knowledge workers in the corporate world.
Let's imagine ChatGPT as this super effective "black box" that is supporting you in your quest to find answers. "Input questions" are your research objectives, and "output" is the synthesis of key findings. This black box will bring a tectonic shift in the efficiency of doing research.
However, it is impossible to figure out the path your questions took through the black box. Your only lever to create a quality output is the quality of the input. Let's break down the input into two parts — What and How.
The "What" — Problem-Solving Hats
Clarity of thought about what you need to ask, in what order, at what depth, with what context is something that AI cannot mimic at this point. As a researcher, you have to put way more effort into understanding the problem and the structure to solve it.
"Searching for the output is not going to be a differentiating skillset; critically thinking about the problem will be one."
The "How" — Prompt Engineering
This is famously known as "prompt engineering." There are tons of guides and videos which teach you how to write good prompts. Mastering this will have an edge in the short term.
My take is that AI will easily be able to learn to self-correct prompts and suggest better ones. The advantage will disappear in the future.
Go back to the basics of problem-solving. Understand the context in entirety, understand the scope of the problem, identify the components, and develop a sense of what is more important.