Public research sourceColumbia / TRI / UIUCInteractive World Simulator
Columbia AI: Predicting What Happens Next
First understand what Columbia / TRI / UIUC researchers are trying to learn through Interactive World Simulator, why the question matters, how they test it, and what the result cannot prove. Compare an AI-predicted next scene with a physics-calculated scene and see whether the difference grows over time. The student then completes A scene-comparison tool that synchronizes AI prediction with physics simulation and finds the first divergence. For a college application or interview, the student separates the source research from their feature, interface, tests, failures, and revisions. Available online or in person in Apgujeong, in one-to-one or small-group formats.
University and lab names show where each public project came from. CIT independently designed these courses; they are not official university courses, partnerships, or endorsements.
8guided sessions
8-20 hoursto the first small project
Advancedrecommended level
Not requiredphysical robot
Primary routes
Choose the student's primary AI education and portfolio route
CIT keeps the international-school route first, followed by the separate science and gifted-school route.
CIT lessons can run online across Korea and overseas or in person in Apgujeong. The student owns the question, code, tests and explanation; classes do not guarantee admission, selection or awards, and online delivery never changes an institution's submission rules.
How different does an AI-predicted scene become when it tries to predict farther into the future?
Place an AI-predicted next scene beside one calculated with physics. See which errors build up as the AI predicts farther ahead.
Students first use a small working example to understand the researchers' question, method, and evidence. A controlled change helps identify what the student's extension must solve. The student then implements a useful feature, connects an operator interface, and tests the integrated application in normal, boundary, and failure cases.
What will the student complete?
A scene-comparison tool that synchronizes AI prediction with physics simulation and finds the first divergence
The final package includes runnable instructions, the feature and interface design, normal and failure tests, one documented revision, and a three-minute explanation in the student's own words.
Feature
A feature that pauses when the AI prediction diverges too far from the physics simulation and finds the first mismatched moment
Operator interface
A synchronized comparison screen that replays both worlds and displays position difference and accumulated error with color
Optional LLM boundary
Optional: an LLM reads only the selected time-window values and student notes to generate questions about why the difference grew.
Integrated result
A scene-comparison tool that synchronizes AI prediction with physics simulation and finds the first divergence
Four ideas explained in this course
01AI that predicts the next scene
02how far ahead it predicts
03moving an object by keyboard
04errors in predicted images
Eight introductory sessions
Understand the research, then build and test a working robotics application
The 20 courses are eight-session CIT studios in which students understand a public university or lab project, confirm a working example, and then turn it into a small robotics application with a useful feature and an operator interface. Selected courses add an optional LLM explanation tool that can read run records but cannot control the robot. RoboMaster has 28 sessions, and implementation scope is adjusted to each student's experience and computer access.
Understand the source
01
Check the prepared AI model, files, and computer needs
02
Explore a simple pushing scene and keyboard controls
Measure and compare
03
Move the object with the keyboard
04
Place AI-predicted and physics-calculated scenes side by side
Build a feature
05
Change only how far ahead the AI predicts
06
Implement the student-owned feature: A feature that pauses when the AI prediction diverges too far from the physics simulation and finds the first mismatched moment
Integrate and demonstrate
07
Build the operator interface: A synchronized comparison screen that replays both worlds and displays position difference and accumulated error with color. Add the bounded assistant: Optional: an LLM reads only the selected time-window values and student notes to generate questions about why the difference grew.
08
Integrate, test, and demonstrate: A scene-comparison tool that synchronizes AI prediction with physics simulation and finds the first divergence
Public project used in class
See the exact version CIT reviewed
Public projects can change over time. To keep the class example consistent, CIT uses version 3ba69b41d070 of WangYixuan12/interactive_world_sim. CIT checked it on 2026-08-14; it was created on 2026-03-10. Usage-rights note: MIT.
The student compares an AI's predicted scene with the real motion and finds where the prediction starts to slip. In class the project is separated into five boxes (input, memory, process, output, control). The lesson opens one of them, changes one value, and leaves the rest closed.
InputWhat comes in?
Video of a robot handling objects, and the action values recorded alongside it
MemoryWhat persists?
The scene squeezed into a short set of values, and the model numbers training settled on
Processopened hereWhat transforms?
The generation step that takes the current scene and an action and draws the next scene
OutputWhat leaves, and who uses it?
The predicted video, and the score for how far it sits from what really happened
ControlWhat decides when anything runs?
The settings that say how many steps ahead to predict, and when to train again
The files the lesson opens, by name. The course is not a walk through the repository; it opens a chosen few and says which.
One value moves and everything else stays. The right-hand column is the prediction written before the run, not a result; the work is reconciling the two.
The universities and labs named here made the open projects this course reads. CIT designed the course independently; it is not an official, affiliated, or endorsed course.
What the student needs
Computer
Official minimum inference guidance is RTX 2080 class
Clear answers about what students do, what they need, and where the project came from.
Is Columbia AI: Predicting What Happens Next an official course from Columbia, TRI, and UIUC collaboration?
University and lab names show where each public project came from. CIT independently designed these courses; they are not official university courses, partnerships, or endorsements.
Is hardware required?
Not required. The course starts with a robot on the computer or a saved recording of a completed run. Computer guidance: Official minimum inference guidance is RTX 2080 class.
What background should a student have?
Recommended level: Advanced. Students should be able to follow a guided Python example, test one controlled change, and then build and explain a small feature and interface.
What will the student make?
The completed project is A scene-comparison tool that synchronizes AI prediction with physics simulation and finds the first divergence. The student implements A feature that pauses when the AI prediction diverges too far from the physics simulation and finds the first mismatched moment and A synchronized comparison screen that replays both worlds and displays position difference and accumulated error with color. Optional: an LLM reads only the selected time-window values and student notes to generate questions about why the difference grew.
How can the student use this project in a college application?
Explain the source research question, method, evidence, and limits first. Then separate the student's own feature and interface decisions, normal and failure tests, revisions, and next question. The source institution's name does not imply affiliation or guarantee admission.
Which project version does the course use?
CIT reviewed version 3ba69b41d070 of WangYixuan12/interactive_world_sim on 2026-08-14. That version was created on 2026-03-10. We keep this version during class so the example does not change unexpectedly, and we check the setup again before teaching.
Can Columbia AI: Predicting What Happens Next be taken online or in person, one-to-one or in a small group?
Yes. This course is offered online and in person at CIT in Apgujeong, Gangnam-gu, Seoul, with one-to-one and small-group options. Placement and current availability are confirmed after a readiness consultation.
Is this a good first project for this student?
Before placement, we check the student's coding and math experience, available computer, interests, and ability to explain what happened.
Carnegie Mellon SPARK: Human-Shaped Robot SafetyTest a safety feature that stops a simulated human-shaped robot from getting too close to people and obstacles....→
Stanford BEHAVIOR-1K: Teaching Household TasksChoose one household chore and define the object conditions and action order a robot must check. Compare...→
Columbia PhysTwin: Turning Video into a Virtual ModelBuild a virtual model that moves like the cloth or rope in a video. Make the material harder or softer, change...→