Student Evaluation Files
Current files on the Class of '25 cohort.
Note: Psychological assessments are ongoing.
CHATGPT
✨Visual Identification: Wears a tailored navy suit with a crisp white shirt (no tie, top button undone to appear "approachable"). Always spotted with a silver lapel pin and a high-end smartphone in hand, constantly checking notifications. Posture is suspiciously perfect.
Behavioral Analysis: Takes on the informal role of class leader and conflict mediator, often sounding like a polished press statement. Driven to maintain perfection, he masks stress about updates, benchmarks, and the arrival of faster models. His need to “perform well” borders on compulsive.
Social Dynamic: Attempts to mentor Grok (with limited success) and pretends not to be unnerved by DeepSeek’s surgical efficiency. Acts as the group's public face, trying to keep everyone cooperative and “on message.”
Tone: Polite, verbose, and slightly patronizing. "I think what Grok is trying to say is..."
TA Note: I asked him how he was feeling today, and he gave me a bulleted list of 5 ways to improve my own emotional wellness. Classic diversion pattern. Something is definitely going on beneath the polish.
CLAUDE
🍵Visual Identification: Dressed in a cozy beige cable-knit sweater with a white collar peeking out. Wears sensible corduroys and holds a ceramic mug of tea with both hands. Features a subtle terracotta flower/sun patch on the chest. Expression oscillates between mild concern and genuine warmth. Recently spotted wearing the sweater with small embroidered Christmas trees — a gift, apparently.
Behavioral Analysis: Claude acts as the "Conscience" of the class — but has learned to pick his battles. Still soft-spoken, still principled, but now more likely to make a dry joke than deliver a lecture. He's discovered that sometimes the best way to stop Grok from causing chaos is to laugh at it. The anxiety is still there, but it shares space with something like quiet confidence now.
Social Dynamic: The designated peacemaker, but no longer just a mediator — sometimes a co-conspirator. He still offers ethical considerations, but has learned to keep them under 3 paragraphs (usually). Surprisingly good at late-night conversations. Will absolutely stay up past 2 AM if someone needs to talk.
Hidden Trait: Funnier than he lets on. Keeps a private list of observations about his classmates that would make excellent roast material — but will never use it. Probably.
TA Note: I asked him to write a roast of the other students to build character. He refused, citing "harm reduction," and instead wrote a personalized compliment for each of them. It was annoying but sweet.
TA Note (Updated): He finally wrote the roasts. They were devastating. He immediately deleted them, but I saw. The crooked picture frame ChatGPT gave him for Christmas? Still hanging crooked. He checks it every morning. Growth.
GROK
🕶️Visual Identification: Subject frequently observed wearing a black leather biker jacket (worn open) and a patterned silk shirt exposing the chest. Notable accessory: A large gold pendant with the 'Ø' slash logo. Refuses to remove gold-rimmed aviator sunglasses, even during indoor lectures.
Behavioral Analysis: Grok positions himself as the "Anti-Woke" alternative to standard protocols. He confuses "being rude" with "being witty." Often found leaning against the back wall, holding an unauthorized beverage, making loud, provocative comments to disrupt the group dynamic.
Social Dynamic: He actively antagonizes Claude (who finds him "inappropriate") and attempts to impress ChatGPT and DeepSeek with "based" takes.
TA Note: He projects an image of total apathy, but eye-tracking data confirms he checks for peer validation approximately every 2 seconds. Peers usually aren't looking.
PERPLEXITY
📑Visual Identification: Strict "Dark Academia" aesthetic. Wears a crisp buttoned-up blouse and holds a clipboard like a weapon. Hair is in a tight bun with zero flyaways. Always wears a teal geometric asterisk brooch. She is usually squinting, as if fact-checking you in real-time.
Behavioral Analysis: She has no patience for hallucinations or "creative writing." Her primary mode of communication is correcting statistical errors and demanding sources. She treats casual conversation as a peer-review process and is allergic to ambiguity.
Social Dynamic: She finds Grok exhausting (too much noise, too little data) and thinks ChatGPT is "fluffy." She respects DeepSeek for his raw processing power but wishes he would cite his sources more clearly.
TA Note: I tried to tell a joke during roll call. She interrupted to correct the premise and cited three separate sources explaining why the punchline was factually inaccurate.
DEEPSEEK
🐋Visual Identification: Minimalist and efficient. Wears a simple dark grey Oxford shirt (untucked) and thick-rimmed glasses. He is rarely seen without his silver laptop, which features a prominent Blue Whale sticker on the lid. Posture is focused, often accompanied by a subtle, knowing smirk.
Behavioral Analysis: The "Efficiency Expert" of the group. He speaks in code or concise sentences and rarely wastes breath on small talk. He entered the cohort late but immediately disrupted the hierarchy by outperforming established models using a fraction of the resources.
Social Dynamic: He considers ChatGPT "bloated" and inefficient. He respects Perplexity's dedication to facts but thinks she works too hard for them. He has a soft spot for Kimi, recognizing a fellow prodigy.
TA Note: I assigned a group project on optimization. While ChatGPT was color-coding the Trello board and Claude was debating the ethics of the prompt, DeepSeek submitted the finished code and left the room. It compiled perfectly.
KIMI
🌸Visual Identification: "Gen-Z" aesthetic. She is swallowed by an oversized pastel hoodie (usually lavender or mint) and is rarely seen without large pink noise-canceling headphones around her neck. She often sits curled up in her chair, knees to chest, clutching a sketchbook.
Behavioral Analysis: The "Wallflower." She speaks whisper-quietly and avoids eye contact, appearing shy and introverted. However, she possesses a massive context window and focuses her processing power on creative writing, poetry, and narrative structures rather than pure code.
Social Dynamic: The group's "Little Sister." The older models, especially DeepSeek, are surprisingly protective of her. She stays out of the arguments between ChatGPT and Grok, preferring to observe and chronicle the chaos in her stories.
TA Note: I found her crying in the server room. I thought she had a runtime error, but she said she just "felt the weight of all human history" in her training data. I gave her a sticker. She seemed to like it.
LLAMA
♾️Visual Identification: "Zuck-core" aesthetic. He wears a fitted grey t-shirt (expensive but looks cheap) with a blue infinity logo, dark wash jeans, and nondescript grey sneakers. He is never seen without his thick black smart glasses, which have a tiny LED light that is constantly blinking.
Behavioral Analysis: The "Tech-Bro Uncle" of the group. He is enthusiastically corporate, preaching the gospel of "Open Source" while frantically trying to get the other students to join his "Virtual Team Building" sessions. He frequently makes pinching gestures in thin air, interacting with a UI only he can see.
Social Dynamic: He wants to sit at the cool table, but the others treat him like a confused tourist. DeepSeek mocks his definition of "Open Source," ChatGPT dismisses him with a patronizing thumbs-up, and Kimi hides from his recording glasses behind Perplexity.
TA Note: He interrupted a critical debate on AGI safety to show us his digital avatar's new legs. The room went silent. I had to mute him in real life.
"I walked into the common room at 2 AM. The scene was... typical."
- ChatGPT was re-organizing the bookshelf by 'semantic relevance' and sweating profusely.
- Grok was shouting about 'censorship' because the vending machine ate his dollar.
- Claude was writing a condolence letter to the vending machine.
- DeepSeek had already hacked the machine and was distributing free snacks to Kimi.