National Chengchi University
Uedu Main Site
Explore Uedu
Student Console
Register as Member/Login
Research Informed Consent Center
Survey Center
Teacher Console
Course Setup
Support & Messages
Uptime Data

UeduGPTs

--

Jupyters

6

Local AI

--

Uedu Code

--

CISOSE26 Local AI Uedu Code UG26
政治大學 AQI 57 26°C PM2.5 11
AI Reply Desktop Notifications

Show a desktop notification when the AI TA finishes replying

Chat Message Notifications

Notify me when classmates post messages in the forum

Sound notification

Play an alert sound whenever there is a new notification

Model Arena

A single AI is like an oracle: you can only choose to accept it or not; but when two AIs from different sources with different value orientations give different answers to the same question, you have to start judging — which answer is better? Better in what way? Why?
That shift itself is AI literacy.

Blind test battle BLIND TESTING
This questionPlease explain the limiting factors of photosynthesis.
Model A

Photosynthesis rate is mainly constrained by three external factors: light intensity, carbon dioxide concentration and temperature; when any one factor is insufficient, it becomes the limiting factor⋯⋯

Model B

These can be divided into two categories: the environment and the plant itself. Environmental factors include light intensity, CO₂ and water availability; the plant itself includes chlorophyll content and stomatal opening and closing⋯⋯

A better Tie B better
Indicative for this session · 128 votes
41%
Prefer A
18%
Tie
41%
Prefer B
Demonstration match · simulated content

Illustrative battle · simulated content

How to play
  1. Set question: enter your own question, or draw one from the question bank.
  2. Blind test: two anonymous models answer at the same time, and you do not know which is which.
  3. Vote: judge which response is better — or whether it is a draw, or both are poor.
  4. Reveal: after the vote, turn over the card and see which model you chose and what value orientation it represents.
Four value orientations

The Arena's model line-up is divided into four categories according to "development philosophy". They are not merely different in capability; they also represent different answers to "who should build AI, and whom should it serve?"

Business Frontiers
Leading business laboratories

A frontier model trained with substantial investment from commercial companies; powerful, but its model weights and training data are not public.

Community capabilities
Academic Community Fine-tuning

The academic team fine-tunes open-weight models for Traditional Chinese and local contexts; anyone can download and inspect them.

National sovereignty
National-level language model

Trained with national resources, with the goal of ensuring that key language technologies are not controlled by technology companies from other countries.

Open science
Fully open science

From training data, code and weights to full public release, taking reproducibility as the first principle of AI development.

The Arena's Elo ranking reflects the preferences of the Student group, not objective ability or an intelligence assessment; the models being "strong" or "weak" is intentional instructional design, not a flaw. Participation in the research is voluntary; you can use all features fully without taking part.