Using AI in Your Writing | What It Does Well and Not So Well

AI is good for preliminary fact-finding, initial proofreading, grammar, initial sources, and potential corrections. It provides an outline based on Wikipedia, similar sources, and other digital articles, but these can contain errors. AI rarely gives the complete picture. It sometimes provides a minor detail, and with new queries, you discover something interesting. However, trust in the AI’s answer might require five or more questions because of contradictions.

I also use AI to trace historical data. When choosing a route on the Underground Railroad, AI would often provide me with an answer, which I would record. Upon further investigation, I found contradictions.

The travel takes place in 1832. During this period, forms and routes of transportation were changing at a rapid pace; sometimes, a week made a difference. I asked about quick steamboats on the dates I needed between Norfolk and Baltimore. AI answered with the company names and routes. It didn’t mention that a cholera epidemic hit by the end of June in 1832, with most of the eastern seaboard shuttered for months. Tens of thousands fell ill, and five to ten percent died. I changed the year to 1834 because 1833 didn’t work either. I came close to having to start over with 1834. My fugitives arrive in New York City on June 26, 1834. The New York anti-abolition riots began on July 7, 1834, and destroyed the very church the fugitives were taking refuge in! This worked. I just got them out of the city faster.

I can’t stress enough that researching complex subjects requires a user with well-developed research skills. Trusting AI is not a good idea unless you are hunting for something simple. For instance, AI taking medical information would be dangerous. Most people struggle to express themselves in a way that provides the information AI functions best with. It’s helpful for research, but only if the input person knows the subject well.

AI’s abilities depend on the quality of the information it processes. The programmer’s integrity, talent, and purpose behind their work are crucial.