by Bill W
One of the challenges of running a one-person operation while trying to cover so many stories is the sheer volume of research needed to produce valid conclusions and responsible reporting.
When someone outside the media wants to understand a subject well, they face the challenge of identifying all the tributaries that feed into the main river. Even asking AI has its limits—I haven’t encountered an AI yet with the bloodhound nose of a seasoned investigative reporter.
So it can be absolutely daunting for a private individual to get a real handle on what they are trying to understand. Writing an article as a reporter requires even more due diligence.
There is one more crucial factor: the media-relations people at companies and organizations have only so much time available. If one publication asking for information has 200 readers while others have thousands, how does a company justify using its limited internal resources?
In other words, if a news site has good intentions but only 100 readers, how does a business or organization conclude that spending three hours answering its questions is a good use of time?
During the early stages of this site, I settled on presenting research questions first to Gemini and Claude. They are going to serve as our first line of “experts”—a form of Consensus AI.
We’ll do as much preliminary research as we can with these systems before contacting real people and showing them what the AIs found. I’ll then ask the human experts to help identify where the AIs may be wrong, where they may be missing critical insights, or where important pieces of information are still absent.
In other words, the AIs get the first crack at the research. Then I follow up with real humans.
I’m going to test this system. I hope it works.