Jeb Bladine: Bringing it home with AI's database research
What’s the big deal with artificial intelligence? Well, for one, at least according to AI itself:
“AI revolutionizes data-based research by automating data collection and analysis, identifying complex patterns in massive datasets that humans cannot, enhancing experiment design and simulation, and enabling predictive modeling to forecast outcomes. It speeds up the research process, reduces errors through automation, helps generate novel hypotheses, and facilitates personalized and more efficient clinical trials by finding participants and creating virtual control arms.”
But don’t miss this warning: Below the question box on our most popular AI platform, it states: “ChatGPT can make mistakes. Check important info.”
In the news business, we experiment daily with AI data-based research, particularly when we find new databases to investigate. One of those is Yamhill County’s online compilation of resolutions and orders from Board of Commissioners actions in the past 10 years.
What to ask AI about county activities? Possibilities abound, but one came to mind because it has been so long since we’ve run any substantive report related to a controversial land use story. So I casually asked ChatGPT to search all county orders and resolutions in those databases and report all those that refer to the Yamhelas Westsider Trail.
The response took what seemed — by AI standards — an excruciatingly long five minutes, partially from following AI’s recommendation to add board meeting minutes to the search. With that, ChatGPT delivered text descriptions and links to one resolution, nine orders and 16 board minutes related to the Yamhelas Trail, plus and organized spreadsheet display.
While waiting for that full response, not to be outdone entirely by ChatGPT, I pulled up a list of 379 articles in our own newspaper database that mention the Yamhelas Westsider Trail. That required a one-second search, providing easy ways to compare past newspaper stories with AI listings of YC Board activities.
This small snippet of research time may seem inconsequential, but consider what’s to come. How long before AI can provide intricate analysis of campaign finance databases that shroud many political contributions? How long before AI fully accesses all Oregon criminal and civil case databases for instant scrutiny and narrations compiled about activities and individuals involved in the justice system?
You may have guessed: ChatGPT says both of those AI services, in varying degrees, already are available.
But back to the trail: I wanted to give ChatGPT one more chance to surprise me with unexpected prowess, so I asked to expand the research into all state agency databases, the Yamhelas Westsider Trail board minutes and available newspaper articles to produce “an appropriate and accurate nonfiction book on the history and possible future of the Yamhelas Westsider Trail.”
I now have that short, six-page treatise, and some new perspectives on how so many things are changing in the worlds of higher education, literature, business, government and daily lives. On the other hand, I’m still remembering that line below the question box that begins all queries to this AI platform:
“ChatGPT can make mistakes. Check important info.”
Jeb Bladine can be reached at jbladine@newsregister.com or 503-687-1223.
Comments