Episode 2

When I first asked Chat GPT to write a technology byline about a new industry standard (a byline I had already written for a client), the result was useless. In my second attempt, I included instructions that were more specific. I clarified what the main point of the byline should be and explicitly stated the need to focus on end-user benefits. As before, I provided the app with the same input I had received from my client and instructed it to produce an article based on “the input below.”

This time, the result was on target in terms of messaging, but badly written. The draft was full of overblown phraseology, e.g. [the standard] “heralds a new era…. pushing the boundaries…. a quantum leap” and so on. This draft also repeated benefits, using identical language, in more than one place.

With heavy editing, this draft might be client-viewable.

Today’s lesson is that learning how to get acceptable results from ChatGPT means understanding what you need to tell it and what it can figure out by itself. I’m not there yet. Stay tuned.