#2 AI Audit
Posted: September 8, 2026 Filed under: Uncategorized Leave a comment »For my AI audit, I chose to have Gemini scan and generate study materials concerning Danny Gonzalez’s video “Speedrunning Conspiracy Theory Rabbit Holes with Drew Gooden”. I have seen this video many times from start to finish, and it is one of my comfort videos. I have a lot of other in-depth hobbies and interests I could have done this assignment about, but I fear that most media I enjoy is very vast and complicated, so even I am not an expert on them. This video is one of the only things I can quote, possibly start to finish.
MIND MAP
PODCAST
VIDEO OVERVIEW (It wouldn’t let me make an infographic 🙁 )
ACCURACY
After reviewing the generated materials, I am not shocked to see that it got most of the facts from the video correct. Though, in the podcast, it did get a fact wrong about the alien story told in the video. In the video, the alien story is told by a guy, supposedly named Clifford, who ran into an alien while fighting in Vietnam, and the alien says his name and warns him against fighting. But the podcast says that the alien is Clifford, which is just plain wrong. Also, in the video overview, it portrays the bent part of the dowsing rods in the wrong orientation in some clips.
I thought it was odd that in the podcast, the “hosts” go on a weird AI tangent at the end; it seems disingenuous for the robot to be talking about the future negative effects of robots.
USEFULNESS
Well, neither the mind map nor the video overview is even about the video, really. They are both made of the conspiracies talked about in the video; actually, they almost display the conspiracies like they are facts, and not humorous like in the actual video. The podcast does, however, speak about the source video as a whole, so I think it was the most effective medium. But I don’t think it helps me understand the source any more than just watching it. I think it was the most effective because it was the only medium that acknowledged that the information from the video is based on conspiracies.
AI WEIRDNESS
The most notable hints given that the materials were AI-generated were the art style and the voice cadence. There isn’t much to say about the mind map because I do believe a real person could make a mind map and have it look like that. The only odd part is that someone would make a mind map about this video at all.
The podcast doesn’t sound like your usual monotone AI voice, but there are still stumbles. The male voice was less authentic-sounding, more robotic, and both voices had the noticeable “AI tone”.
The video overview is OBVIOUSLY AI. The art style used in the video is almost synonymous with AI now. Also, the subtitles included in the video are not accurate to what is being said.
Overall, AI voice and art style are very noticeable to anyone who cares to know the difference. AI prioritizes what is most common in existing media to make something that looks the most “real”. But humans can’t scan through everything on the internet like AI, so humans don’t make all of the most common choices. AI uses the most commercial voice, the most used motion graphics, and the most common art style.
TRUST
Based on this experiment, I would warn people against using AI to summarize or go in depth on comedic YouTube videos, or anything with any level of sarcasm or satire. I can imagine these tools being helpful when consuming actual academic material. But I think it would only be trustworthy with topics that have little discrepencies and are not too complicated. When the AI is faced with differing opinions on one topic, it tends to get confused, which I have noticed in past experiences. Overall, I wouldn’t trust AI any more than you trust a friend who is a little bit smarter than you.