AI Playgrounds v1.9.5: translated replay actions
Vietnamese and Spanish now translate missing replay return and jump-control labels across Pathfinding, Hill Climbing, Wumpus World, K-Means, Q-Learning, and Neural Network. K-Means also gains its translated replay-slider name. Mechanisms, history, classroom drafts, and responsive controls are preserved.
Native checks verify actual visible and accessible labels through both website entry forms and independent offline files, including language round trips and keyboard replay. These translation proposals do not constitute human naturalness or screen-reader validation. Earlier releases and downloads are unchanged.
AI Playgrounds v1.9.5：补全回放操作的翻译
路径搜索、爬山法、Wumpus 世界、K-Means、Q-Learning 和神经网络的回放返回按钮及跳转控件现在补全了越南语和西班牙语标签。K-Means 的回放滑块也补全了翻译名称。算法机制、回放历史、课堂草稿及适应不同屏幕的控件均保留。
原生操作检查覆盖两种网站入口及独立离线文件中的实际可见标签和无障碍名称，包括语言往返切换及键盘回放。这些翻译提案不代表已通过人工语言自然度或屏幕阅读器验证。此前发布的版本及下载文件保持不变。
AI Playgrounds v1.9.4: responsive replay controls
Replay controls now wrap within narrow and tablet panels in Pathfinding, Hill Climbing, Wumpus World, K-Means, and Q-Learning. Selectors in Hill Climbing, Neural Network, Pathfinding, and Wumpus World stay within their panels, including long translated options. Mechanisms, replay history, classroom drafts, and all four learner languages are preserved.
Checks cover all six replay-enabled labs at fifteen viewport sizes in four languages, including history and keyboard return to the latest frame. Directory-link behavior, Labs 13–15 layout alignment, and Wumpus percept icons remain intact. Human learner, educator, and screen-reader validation remains deferred. Earlier releases and downloads are unchanged.
AI Playgrounds v1.9.4：适应不同屏幕的回放控件
路径搜索、爬山法、Wumpus 世界、K-Means 和 Q-Learning 的回放控件现在会在较窄或平板布局的面板内换行。爬山法、神经网络、路径搜索与 Wumpus 世界的选项控件保持在面板内，包括较长的翻译选项。算法机制、回放历史、课堂草稿及四种学习者语言均保留。
检查覆盖六个具有回放功能的实验、十五种视口尺寸及四种语言，包括历史回放与通过键盘返回最新帧。目录链接行为、第 13–15 个实验的统一布局及 Wumpus 感知图标保持不变。学习者、教师及人工屏幕阅读器验证仍暂缓。此前发布的版本及下载文件保持不变。
AI Playgrounds v1.9.3: consistent entry links
The original twelve labs now retain language selection, Guided Challenge, and the correct featured experiment at directory URLs as well as explicit HTML links. Regression checks cover all fifteen labs through both website entry forms and renamed standalone offline files, including language switching and saved classroom responses.
The CNF/SAT sharing fix, Labs 13–15 layout alignment, and Wumpus icon key and responsive percept icons are retained. Human learner, educator, and screen-reader validation remains deferred. Earlier releases and downloads are unchanged.
AI Playgrounds v1.9.3：统一不同入口链接的行为
前十二个实验现在通过目录链接或明确的 HTML 文件链接打开时，均保留语言选择、引导挑战及正确的推荐实验。回归检查覆盖全部十五个实验的两种网站入口以及重命名后的独立离线文件，包括语言切换和已保存的课堂回答。
CNF/SAT 分享修复、第 13–15 个实验的统一布局、Wumpus 图例及随格子缩放的感知图标保持不变。学习者、教师及人工屏幕阅读器验证仍暂缓。此前发布的版本及下载文件保持不变。
AI Playgrounds v1.9.2: CNF/SAT formula-sharing fix
CNF/SAT formula links now preserve the shared formula and any parser diagnostic through startup, including in renamed standalone offline HTML files. Explicit scenario links still take precedence over the native formula, and saved experiment controls apply afterwards. Other playgrounds retain their default experiments.
The Labs 13–15 layout alignment and Wumpus icon key and responsive percept icons are retained. Human learner, educator, and screen-reader validation remains deferred; no WCAG conformance or validated learning outcome is claimed. Earlier releases and their downloads remain unchanged.
AI Playgrounds v1.9.2：修复 CNF/SAT 公式分享
CNF/SAT 分享链接中的公式及其解析错误提示现在会在启动后保留，重命名后的独立离线 HTML 文件也适用。明确指定的情境仍优先于原生分享公式，随后应用已保存的实验控件状态。其他实验保留原有默认情境。
第 13–15 个实验的统一布局、Wumpus 图例及随格子缩放的感知图标保持不变。学习者、教师及人工屏幕阅读器验证仍暂缓；不宣称符合 WCAG 或已验证的学习效果。此前发布的版本及下载文件保持不变。
AI Playgrounds v1.9.1: Labs 13–15 shared page layout
Transformer, Agent, and Game Trees now follow the original twelve playgrounds’ shared header controls, learning tabs, Guided Challenge placement, classroom section, reading width, and collapsed teacher notes. Algorithm-specific controls remain distinct. Language changes preserve classroom drafts and disclosure state. These improvements are included in all three standalone offline downloads.
This patch adds full-page layout regression checks across all fifteen labs. It does not certify every page component or resolve every existing difference in the originals. Human learner, educator, and screen-reader validation remains deferred; no WCAG conformance or validated learning outcome is claimed.
Download standalone HTML files. Previously published releases and their downloads remain unchanged.
AI Playgrounds v1.9.1：第 13–15 个实验的统一页面布局
Transformer、Agent 与博弈树实验现在采用与原有十二个实验一致的页首控件、学习标签页、引导挑战位置、课堂区域、阅读宽度和默认折叠的教师说明。各算法专用控件保留差异。切换语言时保留课堂草稿和折叠状态。这些改进也包含在三个实验的独立离线 HTML 下载中。
本补丁增加覆盖十五个实验的整页布局回归检查，但不表示所有页面组件均已完成审查，也不表示原有实验的所有差异都已解决。学习者、教师与人工屏幕阅读器验证仍暂缓；不宣称符合 WCAG 或已验证的学习效果。
下载独立 HTML 文件。此前发布的版本及其下载文件保持不变。
AI Playgrounds v1.9.0: quality and offline classroom delivery
All fifteen playgrounds retain their mechanisms and four learner languages. This release aligns shared surfaces in Labs 13–15, improves accessibility engineering, adds deterministic cross-lab visual checks, and provides fifteen independently movable offline HTML downloads. Wumpus World adds an icon key and larger, cell-scaled breeze and stench icons; gold is unchanged.
Validation limitation: learner and educator usability studies and human screen-reader testing are deferred. This release does not claim WCAG conformance, validated learning gains, or educator adoption. Automated checks cannot establish those outcomes.
Download individual HTML files or the offline ZIP. Historical releases and their original bytes remain unchanged.
AI Playgrounds v1.9.0：质量改进与离线课堂使用
十五个实验保留原有机制及四种学习者语言。本版本统一第 13–15 个实验的公共界面，改进无障碍工程检查，增加跨实验的确定性视觉回归检查，并提供可独立移动的离线 HTML 文件。Wumpus 世界新增图例，微风与恶臭图标随格子尺寸缩放；金子图标保持不变。
验证限制：本版本暂未开展学习者、教师可用性研究及人工屏幕阅读器测试。不宣称符合 WCAG、已验证的学习效果或教师采用成效。自动检查不能证明这些结果。
下载独立 HTML 文件或离线 ZIP 包（下载说明为英文）。历史版本及其原始文件保持不变。
AI Playgrounds v1.8.1: modern-lab learner parity.
Labs 13–15 now include the same scenario-led learning sequence as the first twelve applets: a featured experiment, five predict–run–explain scenarios, a terminology primer, step-by-step explanation, and teacher prompts in all four learner locales. The release also aligns header controls, removes duplicated state prose, and fixes dark-theme contrast without changing any mechanism result.
AI Playgrounds v1.8.0: algorithm modes and reproducible comparison.
Hill climbing adds seeded repeated-restart benchmarking, K-nearest neighbors adds continuous-target regression, and CNF/SAT adds a bounded first-UIP CDCL learning trace. Existing single-run, classification, and DPLL behavior remains available, with EN/ZH/VI/ES state-preserving controls and deterministic mechanism tests.
AI Playgrounds v1.7.2: modern-lab parity and release assurance.
Labs 13–15 now share the mature suite's navigation, discovery metadata, localized learning support, accessibility-oriented text state, and complete Quick Assign packet actions. The release preserves all fifteen algorithms, curriculum placement, one-assignment-per-lab architecture, four learner locales, and the one-time legacy theme migration.
AI Playgrounds v1.7.1: shared theme behavior across all labs.
Compatibility and consistency patch: Labs 13–15 now use the same persisted theme preference as the original applets, migrate the temporary modern-shell theme key, and preserve the complete v1.7.0 fifteen-lab, all-Quick-Assign, four-language, algorithm, and curriculum boundaries.
AI Playgrounds v1.7.0: Quick Assigns for all fifteen labs.
Completes the Level 1 teacher-assignment layer with one stable 10-15 minute Quick Assign per applet, canonical classroom links, local-only responses, four-locale presentation, and preserved mechanism-first Guided Challenges. No new algorithm or sixteenth lab is added.
AI Playgrounds v1.6.2: public provenance hotfix.
Normalizes current-version labels and analytics provenance across the exact generated Pages artifact after the v1.6.1 post-deployment audit found legacy v1.6.0 markers. No algorithm, curriculum, or assignment behavior changes.
AI Playgrounds v1.6.1: Quick Assign and cross-suite consistency hardening.
Formalized four early-course Quick Assign canaries, repaired landing/search/localization drift, standardized the outer shell of Labs 13–15, and added permanent design-system and final-composition QA contracts.
AI Playgrounds v1.6.0: Game Trees, Minimax, and Alpha-Beta Pruning.AI Playgrounds v1.6.0：博弈树、Minimax 与 Alpha-Beta 剪枝。
Adds the fifteenth lab and thirteenth Foundations lab: a deterministic adversarial-search playground for MIN/MAX backup, exact alpha-beta cutoffs, move-order work comparisons, and prediction-before-reveal challenges.新增第十五个实验、也是第十三个基础课程实验：用确定性博弈树展示 MIN/MAX 回传、精确的 Alpha-Beta 截断、行动顺序对工作量的影响以及先预测后揭示挑战。
AI Playgrounds v1.5.1, HCI and adoption hardening.AI Playgrounds v1.5.1：HCI 与教学采用加固。
Corrects analytics coverage and event/campaign semantics, hardens KNN touch recovery plus Bayesian/Neural responsive behavior, refreshes fourteen-lab product copy, and pilots two ready-to-assign Activity Packs without changing any applet algorithm.修正分析覆盖和事件/活动语义，加强 KNN 触控恢复以及贝叶斯网络/神经网络的响应式布局，更新十四个实验的产品说明，并试点两个可直接布置的活动包；不改变任何 applet 算法。
AI Playgrounds v1.5.0, released August 25, 2026.AI Playgrounds v1.5.0，发布于 2026 年 8 月 25 日。
v1.5 is the engagement-excellence pass: Transformer state continuity and deterministic continuation, a visible and isolated agent runtime sandbox, a DPLL branch/prune tree, and exact Bayesian before/after posterior deltas. Ten already-strong applets are deliberately unchanged rather than decorated.v1.5 是互动体验优化版本：加入 Transformer 状态连续视图与确定性续写、可视且隔离的智能体运行时沙盒、DPLL 分支/剪枝树，以及贝叶斯后验概率的精确前后差异。其余十个已经成熟的 applet 刻意保持不变，而不是添加装饰性效果。
AI Playgrounds v1.4.0, released August 25, 2026.AI Playgrounds v1.4.0，发布于 2026 年 8 月 25 日。
v1.4 keeps fourteen public applets and improves the Transformer and agent-runtime visual journeys, language selectors, curriculum navigation, and visible release provenance before Lab 15.v1.4 保留十四个公开 applet，并在 Lab 15 之前改进 Transformer 与智能体运行时的可视路径、语言选择器、课程导航和可见的版本来源信息。
AI Playgrounds v1.3.0, released August 25, 2026.
v1.3.0 adds Lab 14, Agent Tool Use and Context Protocols, as the fourteenth public applet with EN, ZH, VI, and ES semantic and browser parity, deterministic tool/runtime behavior, provenance-aware context, permission boundaries, and explicit stop decisions.
AI Playgrounds v1.2.0, released August 24, 2026.
v1.2.0 adds Lab 13, Transformer Language Modeling, as the thirteenth public applet with EN, ZH, VI, and ES semantic parity, deterministic reference arithmetic, prediction-before-reveal challenges, and offline browser operation.
Release Notes
AI Playgrounds v1.0.1, released August 8, 2026.
12 bilingual, single-file applets covering search, logic, probability, machine learning, vision, and reinforcement learning.
Five scenarios per applet, a featured comparison, a visual explanation, worksheet prompts, student response packet, and text-state accessibility support.
Printable Teacher Pack, responsive Curriculum Map, one-page Student Lab Packet, public Quality Notes, and release notes.
Each applet runs as one HTML file, works locally without an account, and includes shareable experiments, printable response packets, and bilingual guidance.
Copied links use the current controls as the authoritative state, including manual changes made after applying a scenario.
Applet explanations, dynamic hints, shared controls, landing-page content, and educator materials now switch as complete language experiences.
- Algorithmic tests cover all twelve applets.
- Automated browser checks cover mobile, tablet, and desktop layouts.
- Print layouts and offline operation are included in the public release.
The project provides interactive teaching tools. Claims about learning gains, classroom adoption, or accessibility conformance require separate classroom and user studies.