目标

联合国教科文组织 UNESCO 正在制定面向学生的 AI 能力框架。该框架十分科学，正是我们的 AI 课程想要达到的目标。

这 5 方面目标是：Mindset、Ethics、基础、技术、解决问题。每个方面又分为 3 个层次：理解、应用、创造。下面是各个方面的三个层次的具体目标。

第一个方面：Human-centred mindset

Understand: Critical Reflections on AI

• Critically discusses benefits, limitations, and risks of AI
• Gradually builds and expresses an understanding of human rights, social justice, inclusion, equity…
• Describes the connections between technology and social change

Apply: Safe and Responsible Use

• Develops values on personal usage of AI based on trade-offs
• Takes action to protect personal privacy and security
• Practices safe use of AI
• Identifies risks to human rights, human dignity, equity & inclusion, bias amplification…

Create: Self-actualization in the AI Era- Defines examples of AI in everyday life

• Pursues personal fulfillment in the AI era
• Develops a growth mindset, resilience and persistence
• Develops and defends views on long-term impact of AI on society

第二个方面：Ethics of AI

Understand: Human Agency

• Recognizes AI as human-led
• Understands the critical steps of AI development
• Describes ethical challenges that may arise in AI (bias, privacy, security, transparency, explainability…)

Apply: Ethics by Design

• Assesses the purpose of AI products beyond what is explicitly stated
• Communicates the benefits and potential ethical challenges of AI products engaged
• Explains how the decisions of AI creators impact system outcomes
• Proposes (or implements) modifications to address ethical concerns

Create: AI Citizenship- Understands ethical principles linked to AI: do-no-harm, safety and security, fairness, non-discrimination, the right to privacy, data protection, human oversight and determination, transparency, explainability

• Pursues ways to reduce the environmental costs of AI
• Reflects on existing inequities in developing AI, and root causes
• Advocates for ethical and responsible AI use in society
• Contributes to the co-creation of human- centred and inclusive societies

第三个方面：AI Foundations

Understand: Data, Algorithms, and Models

• Understands how data is collected, processed, and used
• Understands the principles of data ownership & privacy
• Understands the goals, benefits, and limitations of modeling data
• Understands sources of bias (human, data, algorithmic)

Apply: Programming and Data Analysis

• Constructs a database, and performs common operations on it
• Executes pre-processing techniques to ensure robust and fair data
• Analyzes data through models to extract meaning
• Uses programming skills to implement basic algorithms and models

Create: Models and Visual Representations- Describes basic functions and algorithms

• Creates visual representations of data using appropriate techniques
• Creates abstractions of AI systems using flowcharts, diagrams and pseudocode
• Evaluates existing algorithms and models for specific use cases

第四个方面：AI skills

Understand: AI Techniques and Applications

• Understands how AI techniques are applied across categories of AI technologies and domains
• Based on real use cases, evaluates accessibility and the linguistic / cultural diversity of AI
• Describes the challenge of explainability for different types of algorithms and models

Apply: Practical AI Skills

• Uses open-source AI tools and programming libraries
• Debugs pre-existing AI systems
• Designs and implements testing strategies for AI systems
• Implements basic AI techniques for specific applications

Create: Creating AI Products- Knows key AI techniques

• Improves existing open source AI models to fit specific goals
• Evaluates existing AI models for specific problem needs (e.g. across culture or language)
• Skillfully develops AI products using multiple AI tools to solve problems
• Assesses the limitations and risks of personal and peer AI creations

第五个方面：AI for Problem Solving

Understand: Problem Scoping

• Determines which problems can be solved using AI from a technical perspective
• Determines which problems should/n’t be solved using AI from a logical, ethical, or cultural perspective

Apply: Co-design

• Effectively collaborates with peers to plan and design AI projects
• Assesses different AI models against identified problem statements and available inputs
• Assesses the availability of ethical and appropriate data
• Chooses between various possible solutions for the best approach
• Effectively communicates the purpose, use cases and limitations of planned AI projects

Create: Co-creation and Feedback Loops- Identifies problems to support human rights

• Implements AI solutions to solve a problem in collaboration with others
• Provides and receives constructive feedback from peers and community
• Assesses the process from design thinking or other problem-solving methodologies

参考

• 框架初稿，Google Doc
• MIT 6.S062/MAS.S10/MAS.s60 Generative Artificial Intelligence in K12 Education](https://mit-cml.github.io/gen-ai-fall-2023.github.io/) 课程 Slides