ATOMS AI Literacy Group · IAMS, Academia Sinica

AI Competency Excellence

Empowering every IAMS member — researchers, postdocs, graduate students, and administrative staff — to use AI safely and effectively.

Browse Courses Learn About ACE

Three Core Pillars

ACE is built around three themes to take you from understanding AI to truly using it well.

🧭

AI Foundations & Risk Management

Understand what LLMs are, their capabilities and limitations, the hallucination problem, and how to choose trustworthy tools while avoiding common risks.

Learn more →
🔍

Smart Choice of AI Tools

What's the difference between free and enterprise tiers? What are the risks of Chinese AI models? How do you decide which service is safe based on data sensitivity?

Learn more →
💬

Effective AI Interaction

Prompting techniques, how to describe research needs, and letting AI help with literature summaries, manuscript editing, code debugging, and other everyday tasks.

Learn more →

How to Choose a Safe AI Tool?

The same service at different tiers can carry completely different risks. Use these principles to decide quickly.

✅

Relatively Safe

ChatGPT Team / Enterprise, Google Workspace (Gemini) — explicit commitment not to use your inputs for model training. Suitable for general work content.

⚠️ Commitments cannot be verified; sensitive data should still not be entered.

⚠️

Use with Caution

ChatGPT Free, Gemini Personal — your data may be used for training. Avoid entering any unpublished research content or personal information.

🚫

Recommended to Avoid

DeepSeek, Qwen, Kimi, Ernie Bot — subject to China's National Intelligence Law, which grants the government authority to access your data. Responses may be censored or biased.

Rule: Sensitive data (unpublished experimental results, personal information) should never be entered into any online service, regardless of tier.
Contact ACE if you need a local model solution.

Before You Start — Let AI Know You

No matter which AI tool you use, spending five minutes on this first makes all the difference.

😤

Without a settings file

AI doesn't know you. Every chat starts from zero — wrong tone, wrong focus, and anything you told it before is gone the moment you open a new conversation.

✅

With a settings file

AI knows who you are, how you work, and what it must never do — from the very first message. No need to repeat yourself. Just work.

Get the Setup Prompt (free) →

What Can LLMs (Gemini/ChatGPT/Claude) Do for Researchers?

More than a chat tool — LLMs have learned human knowledge and logic through language. Any task that can be described in words, they can help with.

AI Solves Problems Like a Tipsy Genius — and Science May Be Optimizing the Wrong Thing
▶️
AI Solves Problems Like a Tipsy Genius — and Science May Be Optimizing the Wrong Thing
Terence Tao · AI 101 · Chinese
Tao compares AI to a knowledgeable but tipsy collaborator — it throws out ideas at random, and only good guidance turns them into something useful. He also warns science may be optimizing for the wrong things.
Terence Tao Demonstrates How AI Solves Math: A Coin Game
▶️
Terence Tao Demonstrates How AI Solves Math: A Coin Game
Terence Tao · AI 101 · Chinese
In a UCLA talk, Tao uses the Erdős Problems database and a coin-game puzzle to show concretely which steps in research should go to humans and which to AI.
Are Mathematicians Being Worked to Death by AI? "Indigestion" Is Threatening a Millennia-Old Field
▶️
Are Mathematicians Being Worked to Death by AI? "Indigestion" Is Threatening a Millennia-Old Field
Terence Tao · JOHNSON-MATH · Chinese
AI can now write PhD-level proofs in minutes. The real crisis for mathematicians isn't unemployment — it's "indigestion": proofs are piling up faster than the field can review and absorb them.

Further reading: Writing & Text Processing · Knowledge & Reasoning · Workflow Automation · Set Up Your AI Coding Assistant

What Can LLMs (Gemini/ChatGPT/Claude) Do for Everyone?

From official documents, correspondence, and meeting minutes to everyday creative and personal tasks — hand off what AI can handle, and save your time for what truly needs your judgment.

"You Are the Work": Fantasy Author Sanderson's Answer for Creators in the AI Era
▶️

"You Are the Work": Fantasy Author Sanderson's Answer for Creators in the AI Era
Brandon Sanderson · AI 101 · Chinese
Brandon Sanderson — author of Mistborn and holder of Kickstarter's all-time fundraising record — on where creators go from here in the AI era: you are the work, not the output.

Further reading: Administrative Tasks · Set Up Your AI Coding Assistant