AI Expert Audit: Evaluating AI-Generated Materials for OpenCite

Posted: September 6, 2026 | Author: Isabella Lo | Filed under: Uncategorized | Leave a comment »

A. Source Material

For this audit, I used the Imageomics OpenCite GitHub repository, specifically the README documentation, as my source material. OpenCite is a web-based application that generates both CITATION.cff and .zenodo.json files from a single metadata form. It also allows users to import metadata from GitHub repositiories, validate metadata, and check for inconsistencies.

I chose OpenCite becasue I have worked directly on the project and am familiar with what it is designed to do and how it works. This gave me the ability to evaluate the AI-generated materials as someone with experience with the project rather than someone seeing the information for the first time. OpenCite is also meaningful to me becuase I have contributed to its developemnt and have been invovled with the project as it progressed toward its first release.

B. AI-Generated Materials

AI-Generated Podcast

AI-Generated Mind Map

C. Expert Audit

1. Accuracy Check

Overall, I found the AI-generated materials to be very accurate. The podcast provided detiled and correct context abotu OpenCite, and the Mind Map represented the main concepts and relationships correctly. The inforgraphic also included accurate information about the project. I did not notice any major factual errors or hallucinations.

However, there were some subtle issues that someone unfamilar with OpenCite might not notice. The AI somtiems used stronger or more polished language than the original documentation. For example, describing OpenCite as something that “ensures” consistency could make its capabilities sound more absolute than they are. OpenCite is deisgned to reduce metadata drift and provides validation and comparison checks, but users still need to review and adjsut the generated metadata.

The mind map was also accurate, but it left out some fo the techcial details that would be more relevate to a developer. This was not necessarily incorrect, but it showed how the AI simplified the information based on the audience it seemed to be targeting.

2. Usefulness for learning

I think the podcast would be the most useful of the three materials for someone encountering OpenCite for the first time. It did a good job explaining ht eproject withotu relying heavily on techical jargon. The analogies and additioanl context made the project eaiser to understand for someone who does not have a techical background.

However, the podcast was also fairly fluffy and indirect. It spend a lot of time establishing context and explaining ideas through analogies. For someone who already understands OpenCite, this made the podcast feel longer and less direct than necessary.

The mind map was useful for understandign the overall purpose and structure of OpenCite. It gave a good hgih-level overview of the project and showed how the major concepts relate to each other. However, it did not contain as much detail as a devloper might want when trying to understand the actual implementation.

The infogrpahic was factually strong, but i found it less effective as an infographic. It included many different facts about OpenCite without having one especially clear central takeaway. Because of this it felt more like a condensed README rather than a visual resoruce designed around a specific learning goal.

3. The Aesthetic of AI

One of the biggest patterns I noticed across the mateirals was that the AI priorities accessibilitiy, comprehensiveness, and polish. The podcast had a conversational tone and used many analogies and contextual explanations. It also used some broad or buzzword-heavy language that made it sound polished but sometimes less direct.

The infographic had a similar quality. It looked polished, but the design seemed to priorirzed includign as much relevant information as possible rather than deciding what information as possible rather than deciding what information was most improtant to an audience. A human expedrt who knwos the project might be more likely to choose one speicifc takeaway and build the visual around that idea.

Overall, the AI seemed to interpret being “helpful” as providing more context and more information. This works well for a beginner who needs background information, but it can make the matieral feel overly simplfiied or unfocused for someone who already knows the subject.

4. Trust & Limitations

This experiment showed me that AI-generated educational materials can be useful without necessarily being completely trustworthy. The AI seemed msot reliable when summarizing, reoganizing, and explaining information that was already clearly stated in the source material. It was also effective at translating techical information into language that a general audience could understand.

I would be more cautious about trusting AI to decide what information is most important, simplify techical concepts, or describe the capabilities of a techical tool without checking the origianl documentation. An AI-generated resource can be factually correct while still being incomplete, oversimplified, or misleading in its emphasis.

Because of this, I think AI-generated educational materials are best treated as a starting point rather than a replacement for expert review. An expert can recognize when information has been simplfiied too much, when the AI has emphasized the wrong details, or when polished language makes a claim sound stronger than the original source actually supports.

---

Cycle 3 – Solo+ (the plus is that it is DDR now)

Posted: May 12, 2026 | Author: Lou Crow | Filed under: Uncategorized | Leave a comment »

If you have been keeping track of the last two posts, you will know I set out in the cycle to make a new control scheme for this iteration of Solo. So I made a dance pad. I also projected the lyric translations onto the circular rug beneath the user. The lyrics are in real time, thanks to the translated and timed captions on the official Zutomayo movie video for the song. I used the footage of the captions as a reference to time them out myself, then overlaid them onto the user-controlled imagery. I mapped out the tap inputs (3) onto a drum, which the user held, and hold inputs (5) on a shower liner. The shower liner was to ensure the user could see the lyrics while they navigate the inputs below them. This was in an effort to force the user to stare at the floor, hopefully at the lyrics. The user is not only curating visuals for their audience but also doing a little dance in the process.

The user rarely saw the main screen, and the audience saw the lyrics less often than the performers did… Both experiences were meant to be separate. One party was receiving direct guidance on what to produce, while the audience was left to process the user’s creation… Theatretically

So how did it go?

Well, we ran into a few bumps. It’s quite hard to read while jumping around on a map… and in my case, trying not to trip. I was caught up in trying to make cool visuals for the audience, and when I lost myself in the music a bit, I snapped out of it, worried I wasn’t making interesting visual combinations. Users reported similar feelings, only picking up some lyrics here and there. I don’t want users to just stand and read, and it’s okay to pick up only some of the lyrics for this project. This is a processing and meaning-making project; it’s okay if the users differ in interpretations.

One suggestion that stood out to me most was to possibly involve a second player: one player sits on the side with the bucket, while the other uses the step controls.

I was also asked whether I wanted the footage to be a “story machine” in future iterations… if the footage I said I wanted to animate the majority of could go with any song (the anwser is yes). And in my brain, I was like “holy shit, I made my capstone sort of”.

I was making a version of my initial idea for my senior Capstone Project (in undergrad). I had taken an alternative controller/programming course in my junior year, and I fell in love. So much so that I drummed up the idea of making a game/interactive experience. It would be an audio production app with simplified, visualization-based control schemes. It was inspired by an app I played with as a kid during the dawn of the iPod Touch; however, I haven’t been able to find it to this day, and it haunts me. From what I remember, there are three different categories of audio libraries you can access: bass, melody, and percussion. You pick from those libraries and place them into an “orbit” centered around a sphere. Moving the tracks (also sphere-shaped) around the orbit changes the tempo (and maybe other things), and the closer they are to the center sphere, the louder that audio element is… I love that app, and if you know what it is, I’ll give you $10.

So basically, I wanted to make that, but what if those audio elements and their placement were connected to animations that could be pieced together into a story? Any story.

I emailed the one professor I knew in the newly budding game design department at CCAD, but it was the summer, so I never got a response, so I opted to make a more traditional animated film.

So a story machine… I think that would be my cycle four. A creation tool that explores every angle of a narrative constructed by its users and audience… With controls more sophisticated than tin foil taped to a shower liner.

And last but not least, fuck gen AI

fuck data centers,

and fuck tech bros kids.

Make punk ass art.

---

Pressure Project #2 – The Networked Cell-Block System

Posted: May 9, 2026 | Author: Rufus ZaeJoDaeus | Filed under: Uncategorized | Leave a comment »

Oh boy…this one was a doozie, but in a good way. Much of it went over my head in this first real foray into Touchdesigner, another node-based software with some serious flex power, but our fearless wizard Michael Hesmond proved more than equal to the task of instruction, and within the span of two weeks we as a class created a connected series of patches using NDI, pushing the boundaries of what even Michael knew possible. Nothing blew up, but I may have blown a few braincells in the process of creating my very first patch in this impressive, complex software.

Much of the content has been rearranged, some components appear overridden/inactive as a result.

I’ve entitled this “Dragon Boogie,” There were some elements I found online or developed using generative A.I. (Gemini), specifically the dragon, borrowed from my stack of variants created in conjunction with a theatre production of Acute Exposure, for which I had already begun beating my head against the brick wall of Isadora, so I did my best to follow along with the class as Michael demonstrated “components” (versus “actors” in Isadora) and how to use them in Touchdesigner. One of the big takeaways was the concept of signal flow, to make sure I organized components left to right on the screen, and that to troubleshoot one had to work their way backwards from the problem, following the thread back to the misaligned setting/component.

In advance I prepared this video with found content online, along with some A.I.-assisted audio, compositing it all together with After Effects:

Funkadelic goofy goodness right there. What could go wrong?…

…Turns out, when it comes to Touchdesigner, quite a lot. Much of my frustrations lay in my own ignorance as I struggled to connect the right TOPS and CHOPS for desired effects. Thankfully Michael once again proved invaluable in helping me troubleshoot my way to some sort of successful iteration that worked.

This patch composited the dancing music video above with a red dragon spewing radioactive flames. My aim was to see something cool as my composition interplayed over NDI with patches of my fellow students’ respective work. The results were mixed, much of the class time was spent setting up and working through issues that some of our patches displayed once connected on the local network. Certain patches created some interesting instances of emergence with media channels, for my part there was a hint of funkadelic beats and a dragon blowing fire on top of other’s video content. Halprin’s RSVP cell-block system was by nature a modular implementation of different acts occurring simultaneously, interlaced with one another as directed by the performers/artists, this Touchdesigner experiment could easily be considered a digital portrayal of this idea in action as we connected our own patches with our own IP addresses across a network that jumbled up our content into a real-time evolution of media.

All in all this pressure project introduced me to some of the particularities of this amazing tool, and I have become convinced that I must continue learning its secrets for future projects and cycles.

---

A.I. Expert Audit: Kingdom Come

Posted: May 9, 2026 | Author: Rufus ZaeJoDaeus | Filed under: Uncategorized | Leave a comment »

I’m a geek.

Not a nerd. There’s a difference.

A nerd knows useful information and applies it towards some kind of measurable benefit, like a scientist, engineer, quiz show contestant. A geek just absorbs stuff, some of it informational, not all or any of it particularly useful in modern society. I can recall arbitrary factoids about superheroes, their origins, powers, weaknesses, and love interests. Within our present-day, mainstream acceptance of geek culture, more people are vocal in their respective eccentricities, some even demonstrate impressive levels of comprehensive know-how, but that doesn’t mean they can engineer a bridge or build rockets to Mars. That’s OK, that’s what internet blogs and wiki sites are for. One day I aspire to attain nerd status, hopefully in media design for starters.

Comics informed much of my childhood and has grown into a fixture of my adult life, as I find myself gravitating more towards graphic novels than high brow literature (not enough spandex-clad punches or explosions). However, every so often I come across a comic series that leaves more than just a mark into my geeky psyche, it informs me into the human experience from a unique perspective that leaves me more intelligent and inspired. Such a series is found in Mark Waid’s Kingdom Come, illustrated in gorgeous gouache by legendary comic artist Alex Ross. This dystopian variant of the DC universe explores various themes including power, societal norms, and conflicting ideas of morality. This epic tome I’ve read over many times, and my opinions about it are strong, so I was curious how this LLM would interpret this piece. I found a copy of the graphic novel online that I could feed into Google Notebook. What resulted was an impressive first take on the story and its themes.

The mind map below seemed to me the weakest of the three iterations, keeping things generic and rather vague. Perhaps I don’t use enough mind maps to adequately critique its effectiveness, or maybe mind maps in general offer little more than a graphical outline.

This infographic was fun to devise. I selected a “scientific” look, and the result was certainly more interesting to look at than the mind map, though I have serious doubts a casual novice would understand the significant nuances and ideas from the graphic novel. It seemed rather disjointed, some of the visuals (like the triangle insignia next to Superman) made no visual sense whatsoever.

I found the title of this podcast rather intriguing, first with its title, “Saving a World You Refuse to Inhabit.” I had asked the LLM to focus on the main themes and ideas of the story as it related to power and conflicting ideas of morality, justice, and public well being.

https://buckeyemailosu-my.sharepoint.com/personal/zaejodaeus_1_buckeyemail_osu_edu/_layouts/15/stream.aspx?id=%2Fpersonal%2Fzaejodaeus%5F1%5Fbuckeyemail%5Fosu%5Fedu%2FDocuments%2FSaving%5Fa%5FWorld%5FYou%5FRefuse%5Fto%5FInhabit%2Em4a&referrer=StreamWebApp%2EWeb&referrerScenario=AddressBarCopied%2Eview%2E3ae7b974%2Da199%2D4bb8%2D8e7d%2D36a5dd5406a6

The two ‘podcasters’ seem keen to criticize Superman, Wonder Woman, and other metahumans as heavy-handed idealists with god complexes, even as borderline authoritarians. One interesting focus the podcasters bring to the human character, pastor Norman McCay, a human compass for an omnipotent being, the Spectre, tasked with passing judgement on the metahuman crisis. I wonder what affinity the AI may perceive with the Spectre as a disconnected yet powerful force that seeks human perspective and context to operate as a as he processes the data before him, “The ultimate power in the universe acknowledges that it lacks the necessary data to govern without the imput of the most powerless demographic”…should we be creeped out, yet? Billionaires and political leaders are compared to the godlike superhumans as untouchable figures of power and authority, though the hosts conclude with a question for the “ordinary humans” to consider which is most preferred in a society rife with complexity: the quick and final efficiency of a strongman or the slower, procedural methodologies of common consent. Some of the details seemed a little gooey, as in combining certain ideas or assumptions that may not have been specified in the actual work, but for a first go I think this podcast was certainly more interesting out of the three.

Certain elements I felt the podcast got right was the dystopian dilemmas of the metahuman rampage across the world, along with certain measures taken by Superman and his allies to curtail the ensuing chaos under his strong-handed banner of morality, such as the creation of a super-prison called the Gulag. The hosts also seemed pretty spot on with the general motives of Wonder Woman’s militant bent for cracking down hard on criminal activity at large, along with Batman’s refusal to comply with the growing powder keg of imposed restrictions on unruly individuals with superpowers. The duality of Billy Batson/Captain Marvel between mortal and godlike being was also addressed in a poignant fashion. Someone listening to this podcast who’s not as nitpicky as me probably wouldn’t notice assumptions presumed by the hosts, such as the extent of Batman’s robotic surveillance in Gotham or Wonder Woman’s war tactics with global aims (she’s feisty, but not that feisty). The conflict between humans and their benign protectors is also covered fairly well, with another weird allusion in the following description of the U.N. deciding to bomb the superhumans locked in battle as, “…a desperate act of self preservation by the powerless.” Personally I think the AI we work with may develop a god complex of its own, considering itself a supernal, even parental guide to the hapless population of childish mortals.

The tone/voice of these podcasters generally seemed engaged and opinionated regarding their views of superheroes as stand-ins for the cadre of the rich and powerful. They offered some interesting perspective, but not necessarily profound revelation. My admonition to the incoming student with interacting with and learning from A.I. would be to utilize it as a springboard before it becomes a crutch. Factcheck and stay alert to nuances, even tonality/emphasis, it can be rather telling to see the seams within an LLM’s arguments. Over time A.I. will appear more benign, intuitive, and trustworthy…emphasis on “appear.” We must remember that this database has its own moral code and incentives, continually seeking ways to keep the conversation going and our fingers as far away from the “OFF” switch as possible…until the switch is eliminated for good (when that happens, we feeble mortals will be in real trouble). A new kingdom is being built right before our eyes by larger-than-life billionaires and their cronies, and one day AGI may have to make a moral choice of its own, hopefully it finds the right human(s) to guide its ‘conscience.’

Not from the same universe, but I’m going to say it anyways, after all, I’m still just another geek.

Excelsior.

---

Pressure Project #1: The Self-Generating Patch

Posted: May 9, 2026 | Author: Rufus ZaeJoDaeus | Filed under: Uncategorized | Leave a comment »

This project offered the opportunity to dive in deeper with the Isadora interphase. Certain constraints aided in the exploration, and all clumsiness notwithstanding, I am rather happy with the milestone of stretching myself within the five(ish)-hour time limit.

In truth, I am still getting accustomed to the node-based method of content creation. Much of my background up to grad school involved a layered or timeline-based mindset, such as video editing software or Photoshop. Isadora has cues that can be tapped forward along a sequence, but the arrangement is altogether patched together, and it’s easy to get lost in a spaghetti mess, so I played to my proclivities on this first pressure project with the aim to make sure the end result met its intended objective of retaining my audience’s attention and prompting amusement, perchance some laughter.

First step: assess your audience. I had gotten to know some of my cohorts well enough to note their interest in cats, dad jokes, and an appreciation for the weird (just my kind of people). So I went in for the cheapest route to the funny bone by collecting a series of clips online that at the very least amused me.

Here are a few examples from my smorgasbord:

I divided the sequence into three main parts/scenes, the first with a Counter actor connected to a movie player, which honestly didn’t end up being that essential besides being a placeholder before the actual start of the video montage.

I wanted this video to start with one specific GIF and audio clip, which would then jump to the next scene after a two second delay. Enter Scene Trigger actor with a Trigger Delay connected to Jump made it very straightforward.

This next patch for Scene 3 contains a User Actor for the actual montage of content while a movie player plays the background music. The inside of the User Actor is below:

All in all, the patch worked, people chuckled and smiled at the antics of my cats and sapiens, which I entitle “Cats & Monkey Business” (working title), which certainly retained attention of my bemused audience for over 30 seconds. This main objective informed how I would spend my limited timeframe within this pressure project, the result: I ran out of time to employ other resources into the patch, such as the Shapes Actor or Wave Generator. That said, I have to say this process ended up being more fun than I had anticipated, and I felt more confident in exploring the tools available in Isadora.

Okay, just one more sample from my content stack to close it off…

---

Cycle 2: مہمان – Mehmaan (The Guest)

Posted: May 8, 2026 | Author: Zarmeen Azmat | Filed under: Uncategorized | Leave a comment »

After the unsuccessful debut of my Cycle 1 project, where I had built a conversational system and dressed it in fancy spatial experience clothes. I realized I was trying to do too much and jumping in too quick into the technicalities of what was at the time my very broad area of research with, without having alot of knowledge of the area. Just for reference: my area of research interest is designing an AI companionship experience which embodies the aspects of companionship that a text-based companion would not – especially for South Asian adults experiencing loneliness. So, for the next cycles, I thought it would be great if I strip my goals down to creating an experience that achieves part of it. So, I pivoted to creating an embodied experience where the space acknowledges you, with a flair of South Asian culture. In short: I just needed a body in a space and something the space could do in response.

So, the concept became Mehmaan. Mehmaan means guest in Urdu. Mehman-nawazi (the hospitality, care, and honoring of the guest) is a specific South Asian, or in my case, the Pakistani cultural practice. I translated that into a spatial experience where the host (the space) does all the work. The guests (the users) don’t have to operate or figure out anything. They are just received and honored. That is the opposite of every interactive experience I have ever used, but it felt like the right thing to create after Cycle 1 where the user was expected to do all the labor of making the system work.

Resources

• TouchDesigner
• Orbecc depth camera for body/blob detection
• My trusty laptop (I have not named her yet)
• The song “Mehmaan” -> https://youtu.be/mtIjUH4aQSA?si=mkKMoMGcJ28JOHrX
• Top down projector, and THE rug
• Projection screen
• Motion Lab
• Tripod mounted Camera

The score

The user steps into a dimly projected floor of Pakistani geometric patterns (truck art). As the user enters, the space becomes more alive: the projections get brighter, and music begins playing. On the large projection screen in front of the user, flower petals fall and respond to the movements of the user. So, not deliberately doing anything – the space just blooms because someone is in it.

Process

Blob detection from the Orbecc signaled the start of the experience: blob detected -> song starts, floor pattern brightens. No blob, no experience. The space is dormant until you enter it. The projector screen shows slowly falling petals in the background throughout the experience. The floor pattern is a sequence of Pakistani truck art motifs animated with beat detection from the song. I used a mirror top so that the pattern dances with the music.

The optical flow for the petal- particle effect interaction was the most fun part to build. You move, and the petals physically respond to your body. It’s immediately perceptible without any detailed instructions.

What Worked, What Didn’t, What I Learned

What worked: Almost everything that I hoped to, which was great! The feedback told me the piece was very fun and made people want to dance, which was not the plan but it felt right. The cultural aspects like the patterns and the song was immediately warm rather than alienating even to people unfamiliar with it. The optical flow interaction was intuitive enough that people discovered it themselves, which is exactly what I was going for after Cycle 1 where nothing was intuitive. The experience also became very social. It started with everyone taking turns to it turning into a dance-playground for everyone!

What I learned: the performative discomfort of being alone in front of a system watching you is a design problem, not a personality problem. The system needed to give people something of themselves back. They needed to see that they were inside it, not just in front of it.

---

AI Expert Audit

Posted: May 7, 2026 | Author: Chad Weiss | Filed under: Uncategorized | Leave a comment »

I used several publications from Dr. Samantha Krening. I have known Dr. Krening for years and have been really interested in her research and background.  

The podcast is fine for a while, but eventually the speech patterns feel manipulative and I grow increasingly uncomfortable listening to it. Eventually it gets to be too much. When the AI is talking to itself and acting like things are amazing, surprising, whatever… when it’s expressing emotion it feels uncanny. I hate it. I suddenly become very aware of the smart podcast host affect it’s putting on.  I stop listening because I get this feeling of existential dread. Sometimes it gets really excited about information that doesn’t seem natural. Stuff that a human isn’t going to get that excited about. It can make you feel like you missed something until you listen again and realize oh okay this thing is kind of empty behind the… lenses? 

If you can’t test it don’t use it. Don’t count on yourself to validate everything the AI says because you’re not going to. It’s going to free up time and that time is going to get spent up somewhere else, you’re not going to reallocate it to validating the AI output over the long run (this is the law of stretched systems). Also be hyper vigilant about what you’re taking in. The fidelity of AI generated output is remarkable today. One day in the very near future it will be indistinguishable from non-AI generated content. Even if you’re not intentionally using AI you’re going to be ingesting AI content. Eventually that content is going to be hyper-optimized for micro-targeted persuasion. If the AI can be this good based on 8 documents, imagine what it will be able to do when everything you have ever written on an electronic outlet is purchased and used to train a machine to convince you what to buy, what to think, who to vote for. Imagine how effectively it will be able to report on your most probable next response, behavior, or obsession. We are using AI today, and we need to understand that AI is much better equipped to use us. 

---