3 Comments
User's avatar
Lance Cummings's avatar

If you provide the presentation AI with structured notes, its more likely to produce something visual ... at least with Gamma, which is basically a markdown document turned into slides (and several structured layouts available). So if the structure of the note matches one of these structures, the result is much better.

So typically I compose a master prompt on top of my structured knowledge base, then transfer that over to AI, instead of just prompting inside the tool.

Scott Abel's avatar

It wasn't the lack of visuals I noticed, Lance. It was the lack of understanding that AI appeared not to have about the science behind "why" and "when" (under what circumstances) to use specific types of images to buttress an idea or to help the audience "get it" quickly (as opposed to AI's favorite trick or regurgitating everything in your prompt into the image as words).

It's so interesting to see how each tool interprets these things differently. Thanks for sharing your approach and mentioning the Gamma tool. I'll try to make time to check it out.

Lance Cummings's avatar

That makes sense. I'll have to pay attention to this more next time.