# AI for Social Good > A Canadian publication of sourced explainers on artificial intelligence: how it works, who builds it, which rules apply, and its effects on work, health, schools, and the environment. Each article names its sources and the month it was last verified. Pages on health, money, and law are information only and give no personal advice. Readers are in Canada and the United States; each article names the jurisdiction it describes. ## Guides - [AI Basics](https://aiforsocialgood.ca/ai-basics): Core concepts, history, language models, careers and courses, and the limits of what AI can do. - [AI Tools and Industry](https://aiforsocialgood.ca/ai-tools-and-industry): Developers and model families, chatbots, software, vendor claims, business value, and industries from transport to media. - [AI Policy and Ethics](https://aiforsocialgood.ca/ai-policy-and-ethics): Rules and official guidance, copyright, fraud and misuse, ethics frameworks, sector rules, and questions about risk. - [AI and Society](https://aiforsocialgood.ca/ai-for-society): Jobs and labor data, occupations, economic effects, environment and energy, accessibility and schools. ## Topic archives - [AI and Blockchain](https://aiforsocialgood.ca/blockchain): 1 article - [AI and Cryptocurrency](https://aiforsocialgood.ca/cryptocurrency): 3 articles - [AI Education in Canada](https://aiforsocialgood.ca/education): 83 articles - [AI News](https://aiforsocialgood.ca/news): 16 articles - [AI services](https://aiforsocialgood.ca/ai-services): 1 article - [Finance](https://aiforsocialgood.ca/finance): 1 article - [Google DeepMind](https://aiforsocialgood.ca/deepmind): 9 articles - [Health](https://aiforsocialgood.ca/health): 3 articles - [Traveling Salesman Problem](https://aiforsocialgood.ca/tsp): 5 articles - [All articles](https://aiforsocialgood.ca/blog) ## Policies - [Editorial policy](https://aiforsocialgood.ca/editorial-policy) - [Disclaimer](https://aiforsocialgood.ca/disclaimer) ## AI Basics: articles - [How to Tell What Is and Is Not Artificial Intelligence](https://aiforsocialgood.ca/blog/understanding-ai-differentiating-what-is-ai-and-what-is-not-ai): How the OECD defines an AI system, why a fixed-rule program is not AI, and four questions that show what a product really does. - [Real AI vs Fake AI: How to Test What a Product Claims](https://aiforsocialgood.ca/blog/real-ai-vs-fake-ai-unveiling-the-truth-behind-the-hype-and-identifying-genuine-artificial-intelligence): Fake AI means products that do not do what AI marketing says. SEC and FTC cases from 2024 and a question table show how to test a claim. - [What AI Is Not: Five Common Misconceptions Checked](https://aiforsocialgood.ca/blog/artificial-intelligence-understanding-what-it-is-not-and-dispelling-common-misconceptions): AI is not one product, not a fact database, and not automatically neutral. Five common claims checked against OECD, NIST, OpenAI, and a… - [Other Names for AI: What Each Term Covers and Where It Came From](https://aiforsocialgood.ca/blog/a-comprehensive-guide-to-alternative-terms-for-artificial-intelligence): Where the name AI came from and how cybernetics, machine learning, augmenting intellect, and foundation models differ from it. - [AI vs IT: How They Differ in Definition, Testing, and Risk](https://aiforsocialgood.ca/blog/what-sets-ai-and-it-apart-key-differences-explained): Federal law defines IT broadly; NIST defines AI systems by their outputs. How testing, upkeep, and risk differ, and what IT practice still… - [Data Science vs AI: Where Each Term Comes From](https://aiforsocialgood.ca/blog/artificial-intelligence-vs-data-science-exploring-the-superiority-in-the-age-of-technological-advancements): Data science grew out of statistics, AI out of the 1955 Dartmouth proposal. Sources show where the two fields differ and where machine… - [AI vs Data Science: What Official Job Profiles Show](https://aiforsocialgood.ca/blog/artificial-intelligence-vs-data-science-choosing-the-ultimate-tech-superpower): BLS counts 275,600 data scientists and 38,600 research scientists, while Canada's NOC puts AI titles under data scientists. What differs. - [Deep Learning vs AI: Where Neural Networks Fall Short](https://aiforsocialgood.ca/blog/artificial-intelligence-and-deep-learning-an-in-depth-comparison-of-two-revolutionary-technologies): Deep learning is one method inside AI. Benchmarks on tabular data show tree-based models often match or beat it at mid-sized data. - [How AI Differs From Conventional Software, Per NIST](https://aiforsocialgood.ca/blog/why-artificial-intelligence-stands-apart-from-the-rest-unveiling-its-unique-capabilities-and-revolutionary-potential): AI behavior is fitted to data, not written as rules. A worked example and NIST's Appendix B list show what that changes for testing and… - [Is AI a Computer? Software, Hardware, and Model Files](https://aiforsocialgood.ca/blog/is-artificial-intelligence-considered-a-computer-and-how-does-it-work): AI is software and stored numbers that run on ordinary computers; storage arithmetic, Llama 3 training scale, and the Google TPU show how… - [Is AI an Operating System? What Each One Actually Does](https://aiforsocialgood.ca/blog/artificial-intelligence-vs-operating-system-the-battle-for-control-and-efficiency): An operating system shares hardware by fixed rules while an AI model predicts from data. Android Adaptive Battery and MemGPT show where… - [Is AI a Website? Models, Servers, and Web Pages Compared](https://aiforsocialgood.ca/blog/can-artificial-intelligence-be-considered-a-website): Why a chatbot page is not the AI behind it: definitions from W3C and NIST, how a request reaches a model, and how web pages feed AI… - [Artificial vs Automated Intelligence: How Laws Define Each](https://aiforsocialgood.ca/blog/a-comparison-of-artificial-intelligence-and-automated-intelligence): Automation is not always AI, yet rules often cover both. How US federal law, Canada's directive, and a New York City law draw the line. - [What an AI System Is Made Of: Data, Models, and People](https://aiforsocialgood.ca/blog/essential-components-for-unleashing-the-full-potential-of-artificial-intelligence-a-comprehensive-guide): What an AI system needs beyond its algorithm, using NIST's lifecycle: purpose, data, model, testing, and accountable people, with… - [The Five Dimensions Used to Classify an AI System (OECD)](https://aiforsocialgood.ca/blog/exploring-the-vast-dimensions-of-artificial-intelligence-unveiling-the-future-of-technology): The OECD framework describes an AI system along five dimensions. A credit-scoring example and the OECD's own test results show how it… - [What Counts as AGI? Nine Definitions and a Levels Scale](https://aiforsocialgood.ca/blog/creating-the-future-unveiling-the-ultimate-apex-of-human-artificial-intelligence): AGI has no single definition. A 2023 paper compares nine and proposes performance levels from Emerging to Superhuman. - [The Seven Stages of AI: A Popular List, Checked](https://aiforsocialgood.ca/blog/what-are-the-7-stages-of-artificial-intelligence): No standards body defines seven AI stages. The popular list joins Hintze's four types with three capability tiers, and only two items… - [Superintelligence vs Today's AI: What the Difference Means](https://aiforsocialgood.ca/blog/artificial-intelligence-vs-superintelligence-unraveling-the-future-of-technology): Chalmers' three-level scale, Good's 1965 intelligence-explosion argument, its weakest premise, and the October 2025 Statement on… - [Artificial vs Natural Intelligence: What Measurements Show](https://aiforsocialgood.ca/blog/artificial-intelligence-and-natural-intelligence-exploring-the-boundaries-and-potential-of-human-and-machine-intelligence): Neurons are not parameters. A look at brain counts, brain energy use, Chollet's efficiency measure, and a 2024 child-learning study, with… - [AI vs Human Intelligence: What Replacement Evidence Shows](https://aiforsocialgood.ca/blog/artificial-intelligence-vs-human-intelligence-examining-the-pros-and-cons-of-ai-in-replacing-humans): A 758-consultant experiment found AI users faster inside the tool's abilities and less accurate outside them. Labor research explains why… - [Is AI Part of Computer Science? What CS2023 Says](https://aiforsocialgood.ca/blog/artificial-intelligence-vs-computer-science-a-comparison-of-the-best): AI is one of 17 knowledge areas in the ACM, IEEE Computer Society, and AAAI CS2023 curriculum, with 12 of 162 core hours. Still current in… - [Key Topics in AI: How the ACM Classification Maps the Field](https://aiforsocialgood.ca/blog/key-topics-in-artificial-intelligence-understanding-the-significance-and-implications-of-ai): The ACM classification splits AI into nine areas. A table links each to textbook chapters and sorts familiar systems into them. - [Where AI Fits in Data Science: Six Phases and Test Results](https://aiforsocialgood.ca/blog/how-artificial-intelligence-is-revolutionizing-data-science-and-transforming-industries): Machine learning is one of six phases in a data science project. A 2024 benchmark shows how far AI agents get on analysis tasks. - [How Canada's Program Codes Define AI and Data Science](https://aiforsocialgood.ca/blog/artificial-intelligence-versus-data-science-which-is-the-superior-field): Statistics Canada's CIP 2021 puts AI under computer sciences (11.0102) and data science under interdisciplinary studies (30.7001). The… - [Two Cultures of Data Work: Explaining vs Predicting](https://aiforsocialgood.ca/blog/artificial-intelligence-vs-data-science-unveiling-the-ultimate-champion-in-modern-technology): Breiman's 2001 paper splits modeling to explain from modeling to predict. Two of his projects show how each is tested and why a good fit… - [When to Capitalize Artificial Intelligence and AI in Writing](https://aiforsocialgood.ca/blog/does-artificial-intelligence-need-to-be-capitalized): Lowercase as a general term, AI always in capitals, official titles capitalized: examples from Canadian and US government sources. - [Artificial Intelligence as a Noun: Countable or Not?](https://aiforsocialgood.ca/blog/is-artificial-intelligence-countable-or-uncountable-exploring-the-boundaries-of-ai-in-the-modern-world): Dictionaries treat artificial intelligence as uncountable but record a countable sense. When to write AI system, AIs, or an AI. - [Top AI Systems: How Researchers Define the Frontier](https://aiforsocialgood.ca/blog/the-top-artificial-intelligence-systems-in-the-world): Why no single list ranks AI systems, and how Epoch AI's dataset of 3,600 models defines frontier models by training compute. - [The Most Advanced AI: How Laws Measure It by Training Compute](https://aiforsocialgood.ca/blog/what-is-the-most-advanced-form-of-artificial-intelligence-2): How the EU AI Act, its guidelines, and a rescinded US order use training compute (10^23, 10^25, 10^26 operations) to define the most… - [How Deep Learning Works, From Backpropagation to ImageNet](https://aiforsocialgood.ca/blog/the-powerful-and-dynamic-world-of-deep-learning-ai-exploring-its-applications-advancements-and-transformative-potential): Deep learning uses many-layered neural networks that learn features from data. A worked example, the 2012 ImageNet result, Canadian roots,… - [How an Artificial Neural Network Computes, Step by Step](https://aiforsocialgood.ca/blog/an-in-depth-exploration-of-artificial-neural-networks-a-comprehensive-guide-with-practical-examples): A hand-checkable look at one neuron, why XOR needs a hidden layer, and the limits of the universal approximation theorem. - [How a Neural Network Works: One Worked Example, Step by Step](https://aiforsocialgood.ca/blog/an-illustrative-diagram-of-the-working-process-of-an-artificial-neural-network): A hand-checkable neural network example: forward pass, loss, back-propagation, and one weight update, with the 1986 Nature paper and… - [Neural Network Applications: Which Architecture Fits Which Task](https://aiforsocialgood.ca/blog/exploring-the-applications-and-functionality-of-artificial-neural-networks): How network structure is matched to data type, with documented results for images and sequences and a 2022 benchmark where tree models… - [Where Neural Network Research Is Heading: Four Open Lines](https://aiforsocialgood.ca/blog/exploring-futuristic-possibilities-and-innovative-research-ideas-in-the-field-of-artificial-neural-networks): Four documented lines of neural network research, with what each paper reports, the authors' own limits, and an illustrative scaling… - [Learning Rate in Machine Learning: What It Does and How to Set It](https://aiforsocialgood.ca/blog/ai-learning-rate-boost-up-your-ml-models-with-the-optimal-learning-rate): What the learning rate does in model training, how too high or too low rates fail, and how schedules, warmup, and batch size change the… - [Linear Regression in AI: How Least Squares Predicts Outcomes](https://aiforsocialgood.ca/blog/artificial-intelligence-linear-regression-predicting-outcomes-with-data-driven-models): Linear regression fits a line by minimizing squared errors. What NIST and scikit-learn document about its limits, and why Anscombe's… - [What Vectors Do in AI: Embeddings and Similarity](https://aiforsocialgood.ca/blog/understanding-the-role-of-vectors-in-artificial-intelligence-a-comprehensive-guide): How vectors and embeddings represent data in AI, how cosine similarity and the dot product differ, and what 2013 word vector research… - [How Vector Search Works in Retrieval-Augmented AI](https://aiforsocialgood.ca/blog/understanding-the-concept-and-importance-of-vectors-in-artificial-intelligence): How retrieval-augmented AI turns text into vectors, ranks passages by similarity, and why stored embeddings need the same care as the… - [How AI Text Generators Pick Words: Temperature and Top-p](https://aiforsocialgood.ca/blog/the-revolutionary-artificial-intelligence-text-generator-that-is-transforming-content-creation): How an AI text generator picks each word, what temperature and top-p change, and what Google advises before publishing AI-generated text. - [Long Context in AI: Why Models Miss Text in the Middle](https://aiforsocialgood.ca/blog/limited-memory-ai-unlocking-the-potential-of-artificial-intelligence-with-memory-constraints): A 2024 study found accuracy dipped for text mid-prompt; a 2025 test of 18 newer models found length and distractors hurt. Plus a simple… - [More in AI Basics](https://aiforsocialgood.ca/ai-basics) ## AI Tools and Industry: articles - [Anthropic: Company Profile, Structure, and Safety Policies](https://aiforsocialgood.ca/blog/anthropic-ai): Sourced, neutral profile of Anthropic: structure, funding, Claude models, governance, and safety policies, verified October 2026. - [OpenAI Explained: Structure, Ownership, and Oversight](https://aiforsocialgood.ca/blog/openai-empowering-ai-for-a-better-future): Neutral profile of OpenAI: the 2015 nonprofit, the 2019 capped-profit arm, the October 2025 recapitalization, Microsoft terms, and the… - [DeepSeek: Models, Training Costs, Data Storage, and Bans](https://aiforsocialgood.ca/ai-services/deepseek): What DeepSeek has released since 2024, what its reported training cost leaves out, where app data is stored, and which governments… - [Baidu Ernie Bot: Timeline, Models, and Open Weights](https://aiforsocialgood.ca/blog/ernie-bot): Dated facts on Baidu's Ernie Bot: the March 2023 debut, August 2023 public release, ERNIE 4.5 and X1, free access, and open weights. - [Cohere's Models: Command, Embed, Rerank, and Their Licenses](https://aiforsocialgood.ca/blog/cohere): A neutral guide to Cohere's Command, Embed, Rerank, Parse, and Aya models: licenses, hardware needs, and platforms, checked October 2026. - [Grok Open Weights: What xAI Released and Under Which License](https://aiforsocialgood.ca/blog/grok): Grok-1 and Grok 2 weights compared: architecture, download size, hardware, license terms, and what the xai-org listing shows in October… - [Sony AI: Gran Turismo Sophy and the FHIBE Fairness Dataset](https://aiforsocialgood.ca/blog/sony-ai-exploring-the-future-of-artificial-intelligence): Sony AI was established in April 2020. Two Nature papers describe a racing agent and a consent-based image dataset, with limits stated. - [Amazon Nova and Bedrock: Models, Data Terms, and Limits](https://aiforsocialgood.ca/blog/the-revolutionizing-power-of-amazon-artificial-intelligence): Amazon Nova models run through Bedrock: specs, September 2026 retirements, data terms, Region limits, and what is undisclosed, as of… - [Adobe AI: What Firefly Is Trained On and What Terms Say](https://aiforsocialgood.ca/blog/unlocking-the-power-of-adobe-artificial-intelligence-transforming-industries-with-advanced-technologies): Adobe Sensei and Firefly explained: stated training data, terms on customer files, partner models, and what Content Credentials can show. - [GPT-3: The 2020 Paper, Its License, and What Changed Since](https://aiforsocialgood.ca/blog/gpt-3-open-ai-the-revolutionary-language-model-transforming-the-future-of-ai-and-natural-language-processing): The 2020 GPT-3 paper's size, data, and weaknesses, the API and Microsoft license, and what changed through October 2026. - [Was GPT-3 a Chatbot? What 2020 to 2022 Papers Show](https://aiforsocialgood.ca/blog/the-revolutionary-ai-chat-bot-gpt-3-and-its-limitless-possibilities): GPT-3 continued text; Meena and Blender were built for dialogue. Chat was 8.4 percent of sampled prompts in 2022, and GPT-3 base models… - [Google LaMDA Explained: Paper, Limits, and Successors](https://aiforsocialgood.ca/blog/google-ai-lamda-an-introduction-to-the-revolutionary-new-language-model): Google announced LaMDA on May 18, 2021. What its paper reports, the limits the authors stated, the sentience dispute, and how Bard and… - [Lambda and Google AI: LaMDA, AWS Lambda, and Lambda Inc.](https://aiforsocialgood.ca/blog/lambda-google-ai-how-this-new-revolutionary-technology-is-transforming-the-world): LaMDA is Google's 2021 dialogue model. AWS Lambda and the GPU cloud Lambda are unrelated products that share the spelling. - [OpenAI Five: The Dota 2 Bot and What Its Paper Reports](https://aiforsocialgood.ca/blog/the-revolutionary-open-ai-bot-transforming-conversational-ai-forever): OpenAI Five beat Team OG 2-0 on April 13, 2019, but played 17 of 117 heroes and counted abandoned games as wins in its 99.4 percent figure. - [OpenAI Lab: Gym, Gymnasium, and What the Name Refers To](https://aiforsocialgood.ca/blog/openai-lab-transform-the-future-of-ai-with-cutting-edge-research-and-innovation): OpenAI Lab is not an official facility. The phrase points to OpenAI Gym, archived April 8, 2026, and its successor Gymnasium. - [Mercury AI: Mercury AI GmbH and Inception's Mercury Models](https://aiforsocialgood.ca/blog/mercury-ai): Mercury AI GmbH is a German chatbot vendor. Mercury is also a diffusion language model family from Inception. No source links them. - [NLLB-200: Machine Translation Built for 200 Languages](https://aiforsocialgood.ca/blog/discover-the-top-artificial-intelligence-translator-of-2022-the-ultimate-language-solution): The NLLB-200 research model translates among 200 languages. What its 2022 and 2024 papers measured, how the 44 percent gain is defined,… - [Resemble AI: What It Sells and How Detectors Are Tested](https://aiforsocialgood.ca/blog/a-comprehensive-comparison-of-resemble-ai-alternatives-finding-the-perfect-match-for-your-image-and-audio-recognition-needs): Resemble AI sells voice synthesis and deepfake detection. No independent ranking exists, so this explains the company and how detectors… - [ChatGPT vs Claude 2: Why Model Versions and Dates Matter](https://aiforsocialgood.ca/blog/chatgpt-vs-claude): Claude 2 launched in July 2023 and was retired in July 2025. A dated look at model lifecycles and a log for repeatable chatbot comparisons. - [AGI Companies: What They Say and What Is Documented](https://aiforsocialgood.ca/blog/artificial-general-intelligence-companies-advancing-agi-technology-and-shaping-the-future): No agreed test shows any company has built AGI. What three organizations say about their aims, the Seoul safety commitments, and one… - [Frontier Model Forum: Which AI Companies Belong and Why](https://aiforsocialgood.ca/blog/top-ai-companies-changing-the-future-of-technology-and-innovation): Who belongs to the Frontier Model Forum, how a company qualifies, who pays, and what the group published in 2026. - [AI Hackathon Project Ideas With Licensed Data and Clear Limits](https://aiforsocialgood.ca/blog/explore-these-groundbreaking-artificial-intelligence-hackathon-ideas-to-revolutionize-the-tech-industry): Three illustrative AI hackathon ideas built on open government data, the AirNow API, and WCAG 2.2, with the license terms and limits each… - [Local AI on Windows: Windows ML, DirectML, and Foundry Local](https://aiforsocialgood.ca/blog/top-ai-software-for-windows-pc-to-boost-productivity-and-efficiency): How Windows ML, DirectML, and Foundry Local run AI models on a PC, what hardware they need, and what to ask a vendor before installing. - [Open-Source AI Projects: How to Read a License Before Use](https://aiforsocialgood.ca/blog/top-artificial-intelligence-projects-with-source-code): What BSD, Apache 2.0, and MIT licenses ask of users, why code and model weights differ, and a ten-minute checklist for any AI repository. - [Google DeepMind: What the Lab Builds and Who Runs It](https://aiforsocialgood.ca/blog/what-does-google-deepmind-do): Google DeepMind merged DeepMind and Google Brain in 2023. A dated look at its Gemini models, AlphaFold, weather work, safety framework,… - [ChatGPT Explained: GPT Models, Training, Plans, and Limits](https://aiforsocialgood.ca/blog/a-bold-new-era-exploring-the-revolutionary-ai-chatbot-experience-powered-by-gpt-technology): What GPT means, how OpenAI trained ChatGPT, a dated company timeline, plan prices read on October 24, 2026, and the limits OpenAI documents. - [ChatGPT Conversations: Training, Deletion, and Privacy Controls](https://aiforsocialgood.ca/blog/discover-the-powerful-capabilities-of-open-ai-chat-revolutionize-your-conversations-today): How ChatGPT was built and what happens to conversations: training toggle, Temporary Chat, deletion windows, a court order, and Canadian… - [What People Ask ChatGPT: Usage Numbers From an NBER Study](https://aiforsocialgood.ca/blog/chatgpt-chatbot-artificial-intelligences-most-advanced-conversational-assistant): NBER Working Paper 34255 reports 700 million weekly users by July 2025 and that 73 percent of messages in June 2025 were not work-related. - [How to Judge If a ChatGPT-Style Chatbot Is Advanced](https://aiforsocialgood.ca/blog/a-comprehensively-advanced-chatgpt-chatbot-conversing-with-ai-like-never-before): Three public methods for testing chatbots (HELM, Chatbot Arena, MT-bench), what each measures, their documented biases, and a checklist… - [Why Long Chatbot Conversations Go Wrong, Per 2025 Research](https://aiforsocialgood.ca/blog/using-artificial-intelligence-to-have-conversations-unlocking-the-power-of-ai-chatbots): A 2025 study of 15 language models found about a 39 percent average drop when a task is revealed over several turns instead of one message. - [How Chatbots Generate Text, and Why They Make Errors](https://aiforsocialgood.ca/blog/ai-bot-chat-gpt-natural-language-generation-for-enhanced-conversational-experiences): How GPT-style chatbots build replies token by token, what human feedback changed, why answers vary, and why NIST treats confabulation as… - [How Task-Based Conversational AI Works: Slots and Intents](https://aiforsocialgood.ca/blog/unleashing-the-power-of-conversational-ai-exploring-the-leading-companies-in-the-industry): No neutral ranking of conversational AI companies was found. How task-based systems work: frames, slots, intents, dialogue state,… - [How an AI Voice Assistant Is Built, Stage by Stage](https://aiforsocialgood.ca/blog/creating-an-ai-voice-step-by-step-guide-to-build-your-own-artificial-intelligence-voice-assistant): The four stages of a voice assistant, what Whisper and Piper do, how fast they run on small computers, and which voice data rules apply. - [How Alexa Changed From Echo Commands to Alexa+](https://aiforsocialgood.ca/blog/the-evolution-of-alexa-ai-from-voice-assistant-to-smart-home-manager-and-personal-assistant): How Alexa grew from the 2014 Echo and third-party skills to Alexa+, now open to all in the US and Canada, with dated milestones and… - [Alexa+ in the US and Canada: Price, Access, and Voice Data](https://aiforsocialgood.ca/blog/alexa-the-revolutionary-artificial-intelligence-assistant-that-is-transforming-the-way-we-interact-with-technology): What Amazon says about Alexa+ pricing and access in the US and Canada, the 2025 change to voice recording settings, and regulator advice… - [Chai AI Chat App: What Is Verified and What Is Claimed](https://aiforsocialgood.ca/blog/chai-chat-with-ai-friends-unleashing-the-potential-of-artificial-intelligence-in-conversational-platforms): What Chai Research says about its app, users, revenue, and safety, which figures are unaudited, and what public scrutiny of companion… - [Beta Character AI: What the Beta Was and What Replaced It](https://aiforsocialgood.ca/blog/beta-character-ai): Character AI opened as a public beta in September 2022. A timeline from the company's posts, and where beta.Character AI points now. - [Replika's EUR 5 Million GDPR Fine: Findings and Status](https://aiforsocialgood.ca/blog/is-replika-ai-the-future-of-chatbots-in-human-like-conversations): Italy fined Luka Inc., maker of Replika, EUR 5 million in 2025. What was found, what a US complaint alleges, and what is known by October… - [Replika Study: What 1,006 Student Users Reported](https://aiforsocialgood.ca/blog/is-replika-ai-the-future-of-conversational-artificial-intelligence): The main academic study of Replika: 1,006 student users, late 2021 data, self-reported results, and why the 3 percent figure cannot show… - [How Telegram Bots Work and What AI Bots Can See](https://aiforsocialgood.ca/blog/bot-in-telegram): How Telegram bots are registered and run, which limits apply, and what data an AI bot operator can receive, based on Telegram's… - [More in AI Tools and Industry](https://aiforsocialgood.ca/ai-tools-and-industry) ## AI Policy and Ethics: articles - [Canada AI Regulation: Where Federal and Provincial Law Stand](https://aiforsocialgood.ca/blog/canada-ai-regulation): AIDA lapsed in January 2025. A plain summary of Canada's 2026 AI strategy, Bill C-36, provincial rules, and a Canada versus US table. - [EU AI Act Risk Tiers, Bans, Fines, and the 2026 Delay](https://aiforsocialgood.ca/blog/the-potential-dangers-and-risks-posed-by-artificial-intelligence): How the EU AI Act sorts AI into four risk levels, which practices are banned, the fines, and the 2026 change that moved high-risk deadlines. - [Unethical or Useful? What Federal Rules Ask of High-Impact AI](https://aiforsocialgood.ca/blog/artificial-intelligence-unethical-or-a-catalyst-for-transformation): Whether AI is unethical depends on the use. OMB memo M-25-21 sets seven minimum practices for high-impact federal AI, with waivers and… - [Why AI Emotion Recognition Gets People Wrong](https://aiforsocialgood.ca/blog/artificial-intelligence-misinterprets-human-emotions-understanding-the-limitations-of-ai-emotion-recognition-technology): Why facial emotion reading is unreliable, what the EU AI Act bans in workplaces and schools, and what vendors have changed. - [AI Hallucinations in Court Filings: Case Counts and Rules](https://aiforsocialgood.ca/blog/developing-a-deeper-understanding-of-artificial-intelligence-hallucinations-and-their-implications): Court decisions on AI-invented citations now number in the thousands. What Canadian courts and US Rule 11 require of filers, and what… - [Securing AI Systems: What Cyber Agencies Advise in 2026](https://aiforsocialgood.ca/blog/artificial-intelligence-robustly-enhances-security-measures-empowering-companies-to-thwart-modern-cyber-threats-and-safeguard-sensitive-data): Joint cyber guidance on AI from 2023 to May 2026: what CISA, NIST, and Canada's cyber agency publish, what is a draft, and a control… - [What Counts as a Dominant AI Company Under Competition Law](https://aiforsocialgood.ca/blog/the-dominant-global-player-in-artificial-intelligence): Canada tests dominance market by market under the Competition Act; the US Google search ruling reached the Gemini app. What each test says. - [AI and Copyright: Who Owns AI-Generated Content in 2026?](https://aiforsocialgood.ca/blog/artificial-intelligence-and-copyright-examining-the-legal-implications-of-ai-generated-content-in-the-digital-age): What the U.S. Copyright Office, the D.C. Circuit, and Canadian government papers say about AI-generated content, and where training… - [What the U.S. Copyright Office Says About AI Training](https://aiforsocialgood.ca/blog/the-impacts-of-artificial-intelligence-on-intellectual-property): The U.S. Copyright Office's Part 3 report on generative AI training: how it applies the four fair use factors, and why it is advisory and… - [AI-Generated Art Ethics: Training, Credit, and Court Cases](https://aiforsocialgood.ca/blog/the-ethical-and-creative-challenges-of-ai-generated-art-exploring-the-limitations-and-controversies): Consent, credit, and competition in AI art: the Copyright Office Part 3 report, the Andersen lawsuit, and Canada's Suryast case, with… - [Who Owns an AI-Assisted Invention? US and Canadian Patent Rules](https://aiforsocialgood.ca/blog/artificial-intelligence-ownership-unraveling-the-complex-ownership-landscape): AI cannot be named inventor in the US or Canada. The 2025 USPTO notice, the DABUS appeal, and how ownership passes from a human inventor. - [Three AI Policy Disputes: Copyright, Deepfakes, Weapons](https://aiforsocialgood.ca/blog/controversial-artificial-intelligence-debate-topics-exploring-the-ethics-impact-and-future-implications): Three contested AI questions with their legal status as of October 2026: copyright and training data, sexual deepfakes, and autonomous… - [AI Sexual Deepfakes: US TAKE IT DOWN Act and Canada's C-16](https://aiforsocialgood.ca/blog/artificial-intelligences-negative-impact-on-society-unraveling-the-dark-side): US and Canadian laws on AI sexual deepfakes: penalties, the 48-hour removal duty, FTC enforcement since May 2026, and Bill C-16 changes. - [AI Voice Cloning Scams: What Records Show and Which Laws Apply](https://aiforsocialgood.ca/blog/the-dark-side-of-artificial-intelligence-unveiling-the-unsettling-reality-of-ai-hell): The FCC treats AI voices in robocalls as artificial voices, the FBI warns of cloned-voice fraud, and one big grandparent case never… - [Which AI Is Most Dangerous? Facial Recognition Cases Checked](https://aiforsocialgood.ca/blog/discover-the-futuristic-nightmare-most-dangerous-artificial-intelligence-you-should-fear-today): No official ranking of dangerous AI exists. FTC, NIST, court and privacy commissioner records on facial recognition show where documented… - [AI and Cyber Security: Who Gains, Attackers or Defenders?](https://aiforsocialgood.ca/blog/artificial-intelligence-and-the-battle-for-cyber-security-dominance): DARPA's AI Cyber Challenge, Google's Big Sleep, and a June 2026 Canadian warning: what AI changes for attackers and defenders, and what it… - [Why AI Threat Detection Struggles With False Alarms](https://aiforsocialgood.ca/blog/enhancing-security-with-artificial-intelligence-the-future-of-cyber-defense): Why accurate-looking AI security detectors still bury analysts in false alerts, with an illustrative calculation and what NIST says about… - [AI and Nuclear Weapons: What Governments Have Committed To](https://aiforsocialgood.ca/blog/artificial-intelligence-the-looming-threat-to-humanity): US, UK, French and Chinese pledges to keep humans in control of nuclear use, US law, and the UN vote of December 1, 2025, with their limits. - [AI in Cybercrime: What the FBI's 2025 Complaint Data Show](https://aiforsocialgood.ca/blog/artificial-intelligences-revolutionary-impact-on-cyber-security): The FBI's IC3 logged 22,364 AI-related complaints and over USD 893 million in losses in 2025, led by investment scams. What the tag does… - [Paid AI Stock Lists: What SEC and CSA Notices Require](https://aiforsocialgood.ca/blog/top-artificial-intelligence-stocks-for-immediate-investment): What Securities Act Section 17(b), SEC Press Release 2017-79, and CSA Staff Notice 51-356 require when stock articles are paid, and how to… - [Can AI Forecast Crypto Prices? What Two 2026 Studies Found](https://aiforsocialgood.ca/blog/artificial-intelligence-crypto-forecast-unleashing-the-power-of-ai-for-accurate-crypto-predictions): What two May 2026 preprints report about machine learning forecasts of Bitcoin prices, trading costs, and the naive baseline, with a… - [AI Ethical Dilemmas: Where Principles Conflict and Rules Apply](https://aiforsocialgood.ca/blog/artificial-intelligence-and-the-ethical-dilemmas-it-presents-examining-the-moral-implications-of-cutting-edge-technologies): Why AI principles collide, a 2016 fairness trade-off result, and which ethics documents are voluntary and which are binding as of October… - [The Ethical Dilemma of AI: Can Machines Be Given Morals?](https://aiforsocialgood.ca/blog/the-ethical-dilemma-of-artificial-intelligence-ensuring-morality-in-the-age-of-ai): How an unavoidable-harm dilemma is handled in practice: the Moral Machine survey, Germany's 20 driving ethics rules, and UNESCO and NIST… - [What Trustworthy AI Means in NIST, EU, Canadian, and ISO Texts](https://aiforsocialgood.ca/blog/trustworthy-artificial-intelligence-a-review-of-current-trends-challenges-and-solutions-for-ensuring-reliable-ai): Compares how NIST, the European Commission's expert group, Canada's code of conduct, and ISO/IEC 42001 define trustworthy AI, and what… - [Is AI Inherently Unethical? A Documented Case Tested](https://aiforsocialgood.ca/blog/why-most-artificial-intelligence-is-inherently-unethical-and-what-we-can-do-about-it): Is AI inherently unethical? The Dutch childcare benefits case separates design choices from the technology, with Canada's federal rule for… - [What Human-Centered AI Means: Shneiderman, NIST, Stanford](https://aiforsocialgood.ca/blog/the-importance-of-human-centered-ai-in-shaping-the-future): Human-centered AI aims for high human control and high automation together. Shneiderman's framework, NIST's oversight guidance, and… - [Explainable AI Classes: Self-Interpretable, Local, and Global](https://aiforsocialgood.ca/blog/understanding-the-hierarchies-and-classifications-of-explainable-artificial-intelligence-concepts): Explainable AI splits into self-interpretable models and post-hoc methods, local or global. NISTIR 8312 and its four principles explain… - [UNESCO's Generative AI Guidance: A Checklist for Ministries](https://aiforsocialgood.ca/blog/ai-in-education-a-comprehensive-guide-for-policy-makers): UNESCO's September 2023 guidance on generative AI: seven steps for governments, an age 13 minimum for chatbots, and a gap check for… - [What the WHO Says About AI Ethics in Health](https://aiforsocialgood.ca/blog/who-ai-ethics-a-comprehensive-analysis-of-the-ethical-implications-of-artificial-intelligence): WHO's 2021 guidance sets six principles for AI in health; 2024 and 2026 follow-ups cover multi-modal models, liability, and research ethics. - [AI and Christianity: What Evangelical and Anglican Bodies Say](https://aiforsocialgood.ca/blog/the-intersection-of-artificial-intelligence-and-christianity-exploring-the-ethical-and-theological-implications): Evangelical principles from 2019, a 2024 Church of England motion, and Canadian Anglican decisions of 2026 show how Christian bodies… - [AI Detectors in Schools: False Positives and Fair Process](https://aiforsocialgood.ca/education/how-artificial-intelligence-is-having-a-negative-impact-on-education): AI-text detectors misfire, and a score is not proof. Test results, Turnitin's stated error rate, a January 2026 ruling, and schools that… - [Trust Between Students and Teachers When AI Is in the Room](https://aiforsocialgood.ca/blog/understanding-the-importance-of-education-trust-a-key-factor-for-success-in-the-learning-process): School trust research names five facets. A 2026 Harvard report finds AI suspicion running both ways between teachers and students. - [How Education Is Regulated: ESSA in the US, Provinces in Canada](https://aiforsocialgood.ca/blog/understanding-education-regulation-empowering-students-and-ensuring-quality-education-standards): ESSA's testing, rating, and report card rules for U.S. states, the 2025 shift to Labor, and Canada's provincial model under section 93,… - [AI and Workplace Safety: What the Evidence Shows So Far](https://aiforsocialgood.ca/blog/artificial-intelligence-enhances-workplace-safety): A 2025 systematic review found two studies of AI safety tools with injury or illness outcomes. NIOSH risks and Ontario's monitoring policy… - [Crew Rules and Accident Data for Autonomous Ships](https://aiforsocialgood.ca/blog/enhancing-efficiency-and-safety-in-the-maritime-industry-through-artificial-intelligence-technologies): U.S. law presumes a crew on board, Transport Canada requires a manual override, and EMSA warns its accident counts are not risk rates. - [Autonomous Ships and Maritime Law: IMO and Canadian Rules](https://aiforsocialgood.ca/blog/artificial-intelligence-revolutionizes-the-maritime-law-industry): The IMO adopted a non-mandatory MASS Code on May 22, 2026. How it, the collision rules, and Canadian policy treat ships that steer… - [Transport Canada's Rules for Small Autonomous Vessels](https://aiforsocialgood.ca/blog/developing-the-international-legal-framework-for-artificial-intelligence-and-autonomous-shipping): What Transport Canada's Tier 1 policy requires for uncrewed vessels up to 12 meters or 15 gross tonnage, from board approval to 48-hour… - [How US Insurance Regulators Oversee AI Use by Insurers](https://aiforsocialgood.ca/blog/understanding-the-impact-of-artificial-intelligence-in-business-leveraging-cutting-edge-technology-to-drive-success-and-innovation): What the NAIC model bulletin asks of insurers using AI, what examiners can request, and where 29 jurisdictions stand as of October 8, 2026. - [AI and Cannabis in Canada: What the Act, Audit, and Data Show](https://aiforsocialgood.ca/blog/ai-cannabis-in-canada): Cannabis Act promotion rules apply to AI content. An audit of 261 firms, Health Canada reporting, and 2024 survey data show what evidence… - [Is AI a Threat to Humanity? The Evidence on Both Sides](https://aiforsocialgood.ca/blog/the-debate-arguments-against-artificial-intelligence-as-a-threat-to-humanity-and-modern-society): Separates the long-term loss-of-control argument from documented current harms, with the 2026 international report, the skeptics, and… - [More in AI Policy and Ethics](https://aiforsocialgood.ca/ai-policy-and-ethics) ## AI and Society: articles - [Does AI Create Jobs? Three Kinds of Evidence Compared](https://aiforsocialgood.ca/blog/artificial-intelligence-a-catalyst-for-job-opportunities-and-growth): No agency counts AI-created jobs. A census-era history study, a 78 million net forecast, and 2026 payroll data are compared, with limits. - [AI and Unemployment: What 2026 Labor Data Actually Show](https://aiforsocialgood.ca/blog/will-artificial-intelligence-lead-to-widespread-unemployment-and-job-displacement): September 2026 rates were 6.5% in Canada and 4.2% in the U.S. Neither official release records a cause, so AI's role stays open. - [Is AI Cutting Entry-Level Jobs? Canadian and US Evidence](https://aiforsocialgood.ca/blog/the-impact-of-artificial-intelligence-on-employment-challenges-and-opportunities): Total employment grew after 2022 in Canada and the US. The warning sign is narrower: weaker job growth for young workers in AI-exposed… - [How Official Job Projections Account for AI, 2025 to 2035](https://aiforsocialgood.ca/blog/the-impact-of-artificial-intelligence-on-the-future-job-market-how-ai-is-reshaping-the-workforce): BLS projects 3.5 percent job growth to 2035 and assumes AI follows historical patterns; office support is the largest projected decline. - [Will AI Replace Most Jobs? What ILO, IMF, and Yale Data Show](https://aiforsocialgood.ca/blog/will-artificial-intelligence-replace-most-jobs): ILO, IMF, a long-run study of new job titles, and Yale's 2026 tracker compared: what each measures and why none shows most jobs being… - [Job Displacement Data: What Official Surveys Show About AI](https://aiforsocialgood.ca/blog/artificial-intelligence-and-its-impact-on-employment-the-growing-concern-over-job-displacement-and-the-need-for-adaptation): BLS counted 3.3 million displaced long-tenured workers in 2023 to 2025. Official surveys do not name AI as a cause. Notice rules and EI… - [Augmentation or Automation? How the ILO Sorts Jobs by AI Exposure](https://aiforsocialgood.ca/blog/artificial-intelligence-as-the-augmenting-automation-implications-for-employment): ILO 2025 index: one in four workers have some generative AI exposure, 3.3 percent in the highest tier. Exposure is an upper limit, not a… - [AI and Job Loss: Layoff Announcements and Projections, 2026](https://aiforsocialgood.ca/blog/can-artificial-intelligence-replace-human-workers-and-cause-job-loss): AI led stated reasons in 2026 layoff announcements, yet the counts are small beside total layoffs. Challenger, BLS JOLTS, and BLS… - [Will AI Take White-Collar Jobs? What 2026 Entry-Level Data Show](https://aiforsocialgood.ca/blog/will-artificial-intelligence-take-over-white-collar-jobs-and-leave-workers-out-in-the-cold): New York Fed figures for graduates aged 22 to 27: unemployment about 5.6 percent, underemployment 42 percent, with no AI attribution. - [Can AI Replace Scientists? What Labor Data and Studies Show](https://aiforsocialgood.ca/blog/can-ai-replace-scientists-the-future-of-research-and-discovery): BLS projects growth for scientist jobs through 2035. Exposure is not displacement; an AI research system and a large paper study show… - [Will AI Replace Accountants? What Canada's Job Bank Shows](https://aiforsocialgood.ca/blog/will-artificial-intelligence-replace-human-accountants-in-the-future): Job Bank outlooks for accountants differ by province, several pages name AI and online software, and audit signing stays with licensed or… - [Will AI Replace Nurses? What Workforce Data Shows](https://aiforsocialgood.ca/blog/can-artificial-intelligence-replace-nurses): Official data show nurse employment still growing in the US and Canada. AI mostly changes documentation, and Ontario holds the nurse… - [AI in Hiring: What Ontario and New York City Require](https://aiforsocialgood.ca/blog/what-artificial-intelligence-brings-to-human-resource-management-revolutionizing-the-field-of-hrm): Ontario's AI disclosure rule for job postings and New York City's bias audit law compared, with what each requires and what it leaves open. - [Will AI Fully Replace Humans? Where Rules Require a Person](https://aiforsocialgood.ca/blog/why-artificial-intelligence-will-never-fully-replace-humans-exploring-the-vital-role-of-human-intelligence-in-an-ai-driven-world): Nobody can prove AI will never replace humans, but California, Quebec, and federal Canada set specific rules on human review of automated… - [AI Employers: How to Judge Workplace Quality With Evidence](https://aiforsocialgood.ca/blog/top-ai-companies-that-offer-the-best-workplace-environment-for-tech-professionals-and-ai-enthusiasts): Why rankings of AI employers are hard to check, and which Ontario and US employment rules produce facts a candidate can verify. - [Humans Plus AI: What 106 Experiments Found on Teamwork](https://aiforsocialgood.ca/blog/artificial-intelligence-vs-human-intelligence-can-ai-replace-human-intelligence-a-debate-exploring-the-pros-and-cons): Across 106 experiments, human-AI teams beat people alone but trailed the best single performer. Task type and relative skill decided the… - [Human-AI Interaction: The 18 CHI Guidelines and What Tests Found](https://aiforsocialgood.ca/blog/artificial-intelligence-and-human-computer-interaction-enabling-seamless-collaboration-and-enhanced-user-experience): The CHI 2019 paper's 18 human-AI interaction guidelines, grouped by phase, with test results from 49 practitioners across 20 products. - [AI and Handymen: What Repair-Work Data Show About Chatbots](https://aiforsocialgood.ca/blog/ai-and-handyman-the-future-is-here): How general maintenance and repair work scores on Microsoft's chatbot applicability measure, with BLS outlook, Canadian wages, and safety… - [AI and Productivity: What US and Canadian Official Data Show](https://aiforsocialgood.ca/blog/the-economic-impact-of-the-cost-of-artificial-intelligence-on-business-growth-and-innovation-a-comprehensive-analysis): Productivity rose 1.4 percent in the U.S. and 1.0 percent in Canada in Q2 2026, but neither agency attributes it to AI. What the data can… - [AI and Economic Growth: What Estimates and Data Show](https://aiforsocialgood.ca/blog/boosting-economic-growth-the-transformative-role-of-artificial-intelligence): Acemoglu's model gives up to 0.66 percent more productivity over 10 years and an OECD model 0.1 to 0.95 points a year. Official data do… - [AI Grants: Real Funding Programs in Canada and the US](https://aiforsocialgood.ca/blog/the-future-of-artificial-intelligence-grants-accelerating-innovation-and-collaboration): No single AI grant exists. Three named programs in the US and Canada, who they serve, their status in October 2026, and how to spot grant… - [How a Large Language Model Is Built: Inside the Llama 3 Report](https://aiforsocialgood.ca/blog/exploring-the-intricate-process-of-developing-artificial-intelligence-unraveling-the-secrets-behind-the-creation-of-intelligent-machines): Llama 3 shows how a large model is built: data cleaning, scaling laws, 16,000 GPUs, 466 job interruptions in 54 days, and repeated… - [Water Jug Problem in AI: State-Space Search With Python Code](https://aiforsocialgood.ca/blog/water-jug-problem-in-artificial-intelligence-using-python): How to formulate the water jug puzzle as state-space search, test solvability with gcd, and solve it with breadth-first search in Python. - [Open Source AI: What the Definition Requires and Allows](https://aiforsocialgood.ca/blog/advantages-and-applications-of-opensource-ai-empowering-innovation-and-collaboration): What the Open Source AI Definition 1.0 requires, how OLMo, Llama, and scikit-learn differ, and how the EU AI Act treats open licenses. - [How Governments Test OpenAI Models: What Is Documented](https://aiforsocialgood.ca/blog/openai-lab-advancing-artificial-intelligence-through-innovation-and-collaboration): The 2024 NIST agreement with OpenAI, a 2025 security follow-up, a sample evaluation report, and Canada's institute, with what remains… - [Global Partnership on AI: Members, Structure, and Role](https://aiforsocialgood.ca/blog/global-partnership-on-artificial-intelligence-transforming-the-world-through-collaboration-and-innovation): GPAI launched in June 2020, merged with the OECD in July 2024, and has 46 member countries plus the EU. Membership rules, structure, and… - [AI Investment in US GDP Growth: Fed Estimates, 2025 to 2026](https://aiforsocialgood.ca/blog/why-artificial-intelligence-is-reshaping-industries-and-revolutionizing-the-world): Four categories of AI-related investment added 0.97 points to US growth in early 2025. What Fed analyses counted through 2026 and what… - [AI Reshaping Industries? One Treasury Case, Checked](https://aiforsocialgood.ca/blog/the-revolutionary-applications-of-artificial-intelligence-reshaping-industries-and-transforming-the-world): Treasury reported over USD 4 billion prevented or recovered in fiscal 2024 and named machine learning for one of four items. How to read… - [AI and the Environment: Energy Use and Measured Benefits](https://aiforsocialgood.ca/blog/artificial-intelligence-and-sustainability-a-comprehensive-review-of-the-impact-on-our-environment-and-society): Data centers used about 1.5 percent of world electricity in 2024. What IEA, Berkeley Lab, and the CER project, and what AI benefits are… - [AI for the Environment: Spotting Untracked Ships From Space](https://aiforsocialgood.ca/blog/revolutionary-artificial-intelligence-technology-to-save-the-environment): A Nature study used satellites and three deep-learning models and found 72 to 76 percent of industrial fishing vessels were not publicly… - [AI Solar Forecasting: What a Better PV Forecast Is Worth](https://aiforsocialgood.ca/blog/advanced-handbook-on-utilizing-artificial-intelligence-techniques-in-photovoltaic-systems-to-optimize-efficiency-and-performance): Better day-ahead solar forecasts saved USD 87 million a year in one grid model. Where machine learning fits, and what the evidence does… - [AI in Agriculture: Measured Results and Canadian Adoption](https://aiforsocialgood.ca/blog/the-role-of-artificial-intelligence-in-agriculture-advancements-and-applications-in-the-agricultural-sector): Camera-guided spraying cut herbicide 35 to 59 percent in two trials. What Statistics Canada says about drones, GPS, and variable-rate tools. - [AI Crop Maps From Space: How Accurate Are They?](https://aiforsocialgood.ca/blog/exploring-the-impact-of-artificial-intelligence-on-agriculture-revolutionizing-farming-techniques-for-enhanced-productivity-and-sustainability): AAFC targets 85 percent accuracy at 30 m; USDA NASS measured 75.4 percent overall for 2025. How satellite crop maps are made and read. - [AI in Water and Sanitation: What Is Documented and What Is Not](https://aiforsocialgood.ca/blog/revolutionizing-the-water-and-sanitation-sector-with-artificial-intelligence): What peer-reviewed studies show about AI in flood forecasting and leak detection, set against WHO and UNICEF access data and Canadian… - [Ed Roberts and the Independent Living Movement, Explained](https://aiforsocialgood.ca/blog/discovering-ed-roberts-a-pioneer-in-the-disability-rights-movement): Ed Roberts, a polio survivor, founded the Berkeley Center for Independent Living in 1972. A dated biography and what his principles mean… - [AI Tools for Students With Disabilities: Accessibility Rules](https://aiforsocialgood.ca/blog/ai-for-students-with-disabilities): School-provided AI tools must meet WCAG 2.1 AA under the US Title II rule (new dates 2027, 2028) and WCAG 2.0 AA in Ontario. - [Speech Recognition for Disabilities: What One Project Found](https://aiforsocialgood.ca/blog/ai-revolution-in-special-education-transforming-learning-and-inclusion-for-all): The Speech Accessibility Project cut error rates on atypical speech, but it records adults, so classroom results are untested. What the… - [School Districts and AI: Who Decides in the US and Ontario](https://aiforsocialgood.ca/blog/discover-the-purpose-and-benefits-of-an-education-district-in-modern-learning-environments): The US has 12,546 independent school districts and Ontario 72 boards. RAND surveys show AI teacher training spreading faster than written… - [AI in Education in 2023: What Was Said, What Holds in 2026](https://aiforsocialgood.ca/blog/the-role-of-artificial-intelligence-in-education-by-2023-transforming-learning-and-empowering-students): Four dated items from 2023 on AI in schools, with later U.S. orders and Pew teen surveys through 2026, and what still stands in October… ## Optional - [Sitemap index](https://aiforsocialgood.ca/sitemap-index.xml)