When Microsoft CoPilot was asked the question of “Compare the barriers to adoption for the Jacquard loom with the barriers to adoption of the telegraph. Are the barriers to adoption for Project Starline more like those for the Jacquard loom or those for the telegraph?” The Ai companion came up with this: https://456at5.rwu.me/2025/03/04/project-starline/
The summary it creates is surprisingly satisfying answer. From the way it has main bullet points of the faults of each invention for easy understanding, to the conclusion paragraph it has at the end of the assessment that points out the telegraph and Project Starline are more similar is convenient. Each main bullet point also comes with a source to put the readers mind at ease if they doubted it’s though process.
If I were to grade it’s response as a school assignment, I would give it an A- due to the fact it could have been more specific, especially when talking about the costs of each project. What parts were expensive, why were they expensive at the time?
Now that I have graded this Ai’s response, can I have it make a response worth an A? Trying to improve upon it will be hard, but I tried. I first started with CoPilot, and copied the same prompt which it gave me almost the same answer but with no sources. So I added to be more specific and to site sources. The Ai decided to make the sentences longer, but not more specific and also not include citations. I played around with the prompt a bit more with similar results until I tried ChatGPT 4.0 to see what that would come up with. I was having more trouble with that, but not by much. After fiddling back and forth with LLMs, I couldn’t make something that was better than the original.