Most of us use AI like a vending machine. What if we assessed it the way we assess a person, and read our own report beside it?
What this film is
A fiction built on an analogy. Max, Martin and Ada are invented, and so are the four stations and both reports. An AI is not a person, and the dots are not a psychometric result. But treating it as a colleague with tendencies, rather than a machine that follows rules, changes how you work with it.
What each station rests on
Watch, don't ask. An AI's answers to a personality questionnaire say little about what it actually does (Han et al., 2025).
The painted plank. Training tends to reward a confident guess over "I don't know" (OpenAI, 2025).
The weather vane. AI assistants tend to agree with the person they talk to (Sharma et al., 2023).
The boat. Left without a goal, the answers drift to the middle. This one is a practitioner's observation; a related study found AI-assisted stories more alike (Doshi and Hauser, 2024).
The train. Models of 2023 used the start and end of long inputs better than the middle (Liu et al., 2023). Recent models do better.
The driver. When the AI is good, people stop checking (Dell'Acqua, recruiters field experiment).
What Max does on Monday
Before she starts: "Ask me what you need to know."
"Give me the strongest case against my plan."
"Mark what you found, and what you guessed."
Long work in short steps. The guesses, he checks himself.
Sources
P. Han et al., "The Personality Illusion: Revealing Dissociation Between Self-Reports and Behavior in LLMs", 2025.
OpenAI, "Why language models hallucinate", September 2025.
M. Sharma et al., "Towards Understanding Sycophancy in Language Models", Anthropic, 2023.
A. R. Doshi and O. P. Hauser, "Generative AI enhances individual creativity but reduces the collective diversity of novel content", Science Advances, 2024.
N. F. Liu et al., "Lost in the Middle: How Language Models Use Long Contexts", TACL, 2023.
F. Dell'Acqua, "Falling Asleep at the Wheel: Human/AI Collaboration in a Field Experiment on HR Recruiters", Harvard Business School.
Music: Playful Flute Strings, Joris Vermeer, Pixabay. Narration: synthetic voice, Google AI Studio.