Sometimes, before looking forward, it is good to look at where you were. I distinctly recall in November of 2024 that my workflow had significantly changed – ChatGPT had become my go to research tool.

For years, one of my superpowers was to quickly filter through Google results to find specific information. I was usually trying to dig into the technical details of a subject.  To do this, my strategy was the following:

1. Search for the keyword(s)
2. CTRL-Click on the 20 or so most promising links
3. Quickly skim each tab
4. Curate the top 5 articles – down to 3-5 pages
5. Digest and assimilate the details to achieve the solution.

This approach of identify, research, clarify, curate, and solve proved to be very powerful as it allowed me to search many articles quickly. I could regularly find information or answers that would elude colleagues or employees.

My technique was reminiscent of one shared with me while studying for my doctorate in the early 1990s at UC San Diego.  Dr. Alan Sussman, my professor in mathematic optimization explained a key to his success while completing his PhD. He recognized he was not intellectually at the top of his class nor did he have the best scientific intuition. But he did have a skill that set him apart from his peers. He was great at identifying parallel problems in both the literature and using Gopher. He would then apply the math that other’s had already solved to allow him to make progress on his research. Identify, research, clarify curate, and solve.

With the advent of ChatGPT and other AI tools, over the course of 2024, I found the engines capable of finding the information I was looking for faster than my old approach. Seeing a better approach, I slowly started to pivot. My techniques also started to change:

1. Ask the high level question
2. Clarify the question, based on the AI engine answer.
3. Dig deeper with consecutive queries until you uncover the result.

I remembered my thought back in November of 2024 that the future superpower for people trying to find information will no longer be filtering information out of Google. Rather the key will be how well people can ask clarifying questions to the AI to solve the questions that it has gleaned from its training data.

Now, a little more than a year after I first noticed my new default, I’m seeing this 5 step approach to answering questions continuing to be important. But in today’s AI game, there will be other pieces that will also need to be thought about – markdown, memory, and context. But I’ll leave those for another article.