什么是"幻觉"，大语言模型为什么会产生它？
What is a 'hallucination' and why do LLMs produce them?
幻觉指模型生成看似合理但事实错误的内容。因为它按概率预测下一个词而非检索真实事实，缺乏事实校验时就会编造。
A hallucination is plausible but false content. Because LLMs predict the next word by probability rather than retrieving facts, they can fabricate when not grounded in evidence.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Russell & Norvig,《Artificial Intelligence: A Modern Approach》第4版,Pearson,2020
S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed., Pearson, 2020
独立核验依据 · Independent verification: 核验依据 2 ARCHIVE 标识 isbn_9788177583670：Artificial Intelligence: A
#幻觉 #LLM #准确性
什么是过拟合，如何避免？
What is overfitting and how can it be avoided?
过拟合指模型把训练数据的噪声也记住了，导致在新数据上表现差。可用更多数据、正则化、提前停止和交叉验证来缓解。
Overfitting is when a model memorizes training noise and performs poorly on new data. It is reduced with more data, regularization, early stopping, and cross-validation.
难度 专业 · 适合 高中(15-18)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Russell & Norvig,《Artificial Intelligence: A Modern Approach》第4版,Pearson,2020
S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed., Pearson, 2020
独立核验依据 · Independent verification: 核验依据 2 ARCHIVE 标识 isbn_9788177583670：Artificial Intelligence: A
#过拟合 #正则化 #泛化
什么是训练数据和测试数据？
What are training data and test data?
模型先用训练数据学习规律，再用没见过的测试数据检验它是否真的学会，防止只是死记硬背。
A model learns from training data, then is checked on unseen test data to see if it truly generalized rather than just memorized.
难度 进阶 · 适合 初中(12-15)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Russell & Norvig,《Artificial Intelligence: A Modern Approach》第4版,Pearson,2020
S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed., Pearson, 2020
独立核验依据 · Independent verification: 核验依据 2 ARCHIVE 标识 isbn_9788177583670：Artificial Intelligence: A
#训练 #测试 #泛化
什么是强化学习？
What is reinforcement learning?
强化学习让智能体在环境中试错，通过奖励和惩罚来学习能最大化长期回报的策略，AlphaGo 就是典型例子。
Reinforcement learning trains an agent by trial and error using rewards and penalties to learn a policy that maximizes long-term reward; AlphaGo is a famous example.
难度 进阶 · 适合 初中(12-15)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Russell & Norvig,《Artificial Intelligence: A Modern Approach》第4版,Pearson,2020
S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed., Pearson, 2020
独立核验依据 · Independent verification: 核验依据 2 ARCHIVE 标识 isbn_9788177583670：Artificial Intelligence: A
#强化学习 #智能体 #奖励
监督学习和无监督学习有什么区别？
What is the difference between supervised and unsupervised learning?
监督学习使用带正确答案标签的数据来训练；无监督学习面对没有标签的数据，自行发现其中的结构和分组。
Supervised learning trains on data with correct labels; unsupervised learning finds structure and groups in data that has no labels.
难度 进阶 · 适合 初中(12-15)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Russell & Norvig,《Artificial Intelligence: A Modern Approach》第4版,Pearson,2020
S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed., Pearson, 2020
独立核验依据 · Independent verification: 核验依据 2 ARCHIVE 标识 isbn_9788177583670：Artificial Intelligence: A
#监督 #无监督 #学习
什么是神经网络？
What is a neural network?
神经网络由许多层相连的处理单元（神经元）组成，通过调整连接权重从数据中学习复杂关系，是深度学习的基础。
A neural network is layers of connected units (neurons) that learn complex relationships by adjusting connection weights, forming the basis of deep learning.
难度 进阶 · 适合 初中(12-15)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Russell & Norvig,《Artificial Intelligence: A Modern Approach》第4版,Pearson,2020
S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed., Pearson, 2020
独立核验依据 · Independent verification: 核验依据 2 ARCHIVE 标识 isbn_9788177583670：Artificial Intelligence: A
#神经网络 #深度学习
什么是训练一个AI模型？
What does it mean to train an AI model?
训练就是让模型反复调整内部参数，使它对训练数据的预测越来越准，就像学生不断练习改错题。
Training repeatedly adjusts a model's internal parameters so its predictions on the data improve, like a student practicing and fixing mistakes.
难度 进阶 · 适合 初中(12-15)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Russell & Norvig,《Artificial Intelligence: A Modern Approach》第4版,Pearson,2020
S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed., Pearson, 2020
独立核验依据 · Independent verification: 核验依据 2 ARCHIVE 标识 isbn_9788177583670：Artificial Intelligence: A
#训练 #模型 #参数
什么是机器学习？
What is machine learning?
机器学习让计算机从大量数据中发现规律，而不是由人把每条规则都写死，学会后就能对新数据做预测。
Machine learning lets computers find patterns from lots of data instead of being given every rule, then make predictions on new data.
难度 入门 · 适合 小学高段(9-12)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Russell & Norvig,《Artificial Intelligence: A Modern Approach》第4版,Pearson,2020
S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed., Pearson, 2020
独立核验依据 · Independent verification: 核验依据 2 ARCHIVE 标识 isbn_9788177583670：Artificial Intelligence: A
#机器学习 #数据
电脑是怎么"学"东西的？
How does a computer 'learn' things?
电脑看到很多例子后，会慢慢找出其中的规律，下次遇到新的就能猜对，这就像小朋友多看几次就学会认猫。
After seeing many examples, a computer finds the pattern and can guess right on new ones—like a child learning to recognize cats after a few tries.
难度 入门 · 适合 小学低段(6-8)
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Russell & Norvig,《Artificial Intelligence: A Modern Approach》第4版,Pearson,2020
S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed., Pearson, 2020
独立核验依据 · Independent verification: 核验依据 2 ARCHIVE 标识 isbn_9788177583670：Artificial Intelligence: A
#学习 #例子 #AI
什么是人工智能？
What is artificial intelligence?
人工智能是让机器模仿人类智能去感知、学习、推理和决策的技术，比如语音助手和推荐系统。
Artificial intelligence is technology that lets machines imitate human intelligence to perceive, learn, reason, and decide, like voice assistants and recommender systems.
难度 入门 · 适合 大众
CC BY 4.0 · 已核验 · Verified 2026-08-23 · 2 条独立证据
来源 · Source
Russell & Norvig,《Artificial Intelligence: A Modern Approach》第4版,Pearson,2020
S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 4th ed., Pearson, 2020
独立核验依据 · Independent verification: 核验依据 2 ARCHIVE 标识 isbn_9788177583670：Artificial Intelligence: A
#人工智能 #基础