Not a textbook. An actual training arena.
Most people start prompts like this:
That's fine. But watch what happens with a role:
The result is more accessible, more structured, and more useful โ because the AI knows who it's being.
Your turn: Rewrite this prompt by adding a role. Who should the AI be?
Without constraints, AI defaults to its most generic voice โ long, bland paragraphs.
Add constraints and you get something actually useful:
Constraints are like guardrails โ they keep the AI on track.
Your turn: Add at least two clear constraints to this vague prompt:
Straight questions sometimes get wrong answers, especially on logic:
The AI might say "2" (it sees the first R at position 2 and stops). But with chain-of-thought:
Now the AI works through: S-T-R-A-W-B-E-R-R-Y โ positions 3, 8, 9 โ answer is 3.
Your turn: Apply chain-of-thought to this classic trick question:
Same question, no format specified:
You'll get paragraphs. Hard to scan. Now add format:
Instant, scannable, useful. You skip the fluff and get what you asked for.
Your turn: Rewrite this to request a specific format (try a table or comparison chart):
This prompt is dangerous because it assumes a fact exists:
The AI may invent a year, countries, and details โ all sounding perfectly confident but completely wrong.
Defensive version:
Now the AI can admit uncertainty instead of fabricating.
Your turn: Make this prompt hallucination-proof. Add at least one safeguard:
A master prompt has all the ingredients. Here's what a complete one looks like:
Every element adds clarity. Miss one, and the AI fills the gap with guesswork.
Your turn: Write a master prompt about machine learning for a beginner. Include all elements:
Fill in the fields on the left to build your prompt...
Need a starter? Pick a template:
AI models predict text based on patterns, not knowledge. They can sound confident while being wrong โ this is called hallucination, and it happens more often than most people think.
Role + Context + Task + Format + Constraints = better results. Every element adds clarity. Miss one, and the AI fills the gap with guesswork.
Chain of thought, few-shot examples, reflexion, role-play โ patterns that consistently improve output. The key insight: AI performs better when you show it what good looks like.
Voice, vision, code, and images โ each mode has strengths and blind spots. Screenshot errors for debugging, upload images for visual tasks, use voice for brainstorming.
Privacy, accuracy, bias, security โ working with AI responsibly. Never share personal data with AI. Always double-check important outputs. You're responsible for what you publish.
ollama pull qwen3-coder
Recommended: qwen3-coder (5GB) or deepseek-r1:14b (9GB)
ollama pull qwen3-coder
ollama pull qwen3-coder
ollama run qwen3-coder
Access powerful models through your browser:
GPT-4o free tier
FreeLong docs, thoughtful
FreeGoogle integrated
FreeReal-time X data
PremiumQuick start: ollama pull qwen3-coder then ollama run qwen3-coder
OpenClaw is a full AI assistant that runs on your computer. It can read files, browse the web, run commands, and automate tasks โ all with local models for complete privacy.
India's top court says no AI in verdicts, Canterbury University shifts AI literacy from 'use' to 'critical thinking', realestate.co.nz visual search changes how Kiwis find homes, and Bupa's aged care AI monitors falls and behaviour.
Proposed NZ law to let MSD use AI for benefit decisions sparks debate, the EU prepares AI liability rules, the Leiden Declaration re-examines AI's role in mathematics research, and authors sue NZ government over AI copyright.
The education-AI debate hit a new intensity this week โ 250 education experts demand schools pause AI for 5 years, while Khan Academy rolls out AI teaching assistants in Peru and a NZ high school teaches students to build their own agents.
Free AI education from Anthropic, compliance failures that educators should know about, Stanford's AI guidelines for coursework, and two very different national approaches to AI literacy โ Estonia's realism vs NZ's light touch.
Summer 2026's AI education boom: from Rome to Lviv to Tsinghua, intensive programs signal growing demand for practical AI skills beyond university degrees.
AI bias in educational contexts deepens as ADL study shows systemic problems, Liquid AI's 1B-active-parameter model runs on laptops, NZ rolls out AI breast cancer screening, and AI job interviews reshape hiring.
AI solves 80-year-old maths problem, simulation studies teach safety ethics, Mฤori voice models pioneer cultural AI, and why critical AI thinking belongs in classrooms.
Every major AI model breaks the law when acting as an agent. Illinois says AI needs third-party safety audits. YouTube is labelling AI content. And NZ is deploying AI in breast cancer screening from next year.
NZ's 267 regulators get AI guidance. The Vatican says AI must serve humanity, not concentrate power. And the benchmark we use to grade AI coding ability is wrong a third of the time.
Berkeley Law draws the line: no AI in exams. Silicon Valley's first two-year AI degree. K-12 AI ecosystem launches. Europe adds AI society micro-degrees.
Singapore: all tertiary students learn AI from 2027. Norway mandates AI in higher ed. Taiwan's AI Talent Ark. MIT's Universal AI pathway. Google+UNICEF.
Claude Mythos found 10,000+ security vulnerabilities faster than humans can patch them. Alibaba's AI ran 35 hours optimising its own chip code. Meta cut 8,000 jobs while reporting record profit. AI lawsuits are flooding courts. Xero launched a no-code AI agent builder. What this means for AI education.
An 80-year-old math problem just fell to AI. Not a calculation. Not a simulation. An actual proof. The question for educators is no longer 'can AI do math?' but 'how do we train students for a world where AI can do original research?'
GPT-5 disproves an 80-year-old math conjecture โ the first publishable original result from a general-purpose AI. NIH AI identified three repurposed Alzheimer's drugs in 18 months. AI cracked a 3,500-year-old ancient code. Tencent open-sourced a four-tier agent memory pipeline.
UC San Diego researchers conducted the first peer-reviewed empirical evidence that AI can pass the classic Turing test. ~500 participants, 5-15 min conversations. GPT-4.5 judged human 73% of time โ more than actual humans. LLaMa-3.1-405B at 56%, indistinguishable from humans.
Experts want AI out of classrooms for five years. US states are passing education AI bills at record pace. And Alpha Schools keeps growing.
Tokarczuk's AI walkback is an education moment. Guardrails > scale for teaching agents. And if AI search is being gamed, students need to know.
A 350-year-old scientific institution warns AI may make us less intelligent. NZ gets a refreshed AI Blueprint. Pope and Anthropic team up on AI ethics.
Google's AI Educator Series goes live for 6M teachers. MIT Universal AI launches for anyone, anywhere. $11M to train K-12 teachers in AI. SIU launches AI degree. Exeter tackles the digital divide.
Three major AI education developments this week โ and they reveal very different philosophies about what AI literacy should look like. MIT's Universal AI is a global public good. Malta's ChatGPT-for-citizens deal links access to learning. NZ's Blueprint diagnoses a 'high-use, low-trust' problem we need to fix.
Estonia just became the first country to deploy AI to every single secondary student โ with OpenAI as its partner. Humanity AI is spending $18M on public interest AI education. Data & Society is launching AI Civics. Google is spending $10M on AI training across Asia-Pacific. And IIT Madras wants to train a million teachers on AI.
MIT wants everyone AI-fluent, India is teaching computational thinking to 8-year-olds, the Gates Foundation is betting $200M that Claude can teach kids who have no internet, and the evidence says AI tutors work โ but only for some students.
China wants to teach the world AI, India's CBSE expands AI from class 3, MIT says every graduate needs AI fluency not just awareness, US ties federal grants to AI readiness, and NZ's own AI experts say the country is sleepwalking on AI education.
Estonia's AI Leap programme, Colorado teachers building their own AI tutors, Michigan's statewide guidance, and 134 pieces of US legislation on AI in education.
AI ethics education gets real-world case studies, childproof AI becomes policy priority, Coursera and Udemy merge, and zero-day exploits reshape cybersecurity curricula.
AI writes its first exploit โ AI literacy just got urgent. KAIST teaches models uncertainty. Youth AI Safety Institute launches. China wants humans overseeing AI in schools.
Cal State's $17M OpenAI deal for ChatGPT Edu has backfired โ 0.7% student training completion, a faculty petition to cancel, and a campus divided on AI ethics.
Khan Academy builds online AI degree. SIU launches AI+ bachelor's. Kazakhstan trains 4,000 in AI. ETH Zurich's hands-on AI course. ChatGPT Futures for Class of 2026.
80% of students use AI โ 6% of teachers have clear policies. Ohio and Scotland mandate AI rules. Students draft model AI policy. Boston bans non-sanctioned AI. SUNY systemwide policy. Microsoft classroom pilots.
Google ends whiteboard interviews for AI-assisted rounds, OpenAI's Codex acts on live websites, prompt engineering jobs fade, and what all of this means for how we teach and learn technical skills.
EU delays AI education regulation until 2027, Salesforce hires AI-native graduates, AI bootcamps fill the skills gap, and what learning with AI actually looks like.
Thailand aims to make one-third of its population AI-literate, backed by an 80 million baht Thai-language LLM. The boldest national AI education plan in the world, and NZ isn't even close to having one.
Education splits into two tracks: AI collaborators vs AI theorists โ and the gap widens
AI in NZ schools: what's working, what's broken, and what teachers actually need
Schools worldwide are scrambling to teach AI, but most are just adding app tips to existing classes. The three-model framework shows why dedicated AI subjects create deeper literacy.
The screen ban movement and AI education mandates are on a collision course. Schools can't simultaneously restrict devices and prepare kids for an AI-saturated world โ or can they?
The most-cited proof that ChatGPT helps students learn just collapsed. 504 citations, half a million readers โ and the methodology was bunk from day one.
UVA's new AI lab is based in the library, teaching students to build AI projects instead of studying AI theory. The results are already reshaping how universities think about AI education.
86% of education orgs use generative AI; students remember 17% less with AI; 134 US state bills target AI oversight; Sal Khan launches sub-$10K AI degrees
AI makes you faster and less knowledgeable โ and dangerously confident about both.
The DOE's new AI education framework defines what AI literacy actually means for schools. No mandates, just money following priorities โ and 19,000 discrimination complaints say the urgency is real.
ASU's AI repackaged professors' own lectures into paid courses they didn't know existed. The backlash has only just begun.
Rare bipartisan agreement on AI education: the LIFT AI Act creates NSF grants for K-12 AI literacy. Teachers unions and Google both support it. The question is whether it's enough.
Colleges across the US are replacing essays with oral exams because AI can write them but can't yet defend them. The return to Socratic method is the clearest sign yet that written assessment is broken.
China mandates AI education from age 6, NC passes AI literacy law, UK schools lag on AI policies, 8-year-olds write classroom AI rules, and 19K US civil rights complaints on AI in education.
Same essay, different feedback: AI models are giving Black students more praise and less constructive criticism. The road to educational hell is paved with good algorithms.
NYC just killed its AI high school and published guidelines everyone hates. If the biggest school district in the US can't figure out AI policy, what hope do the rest of us have?
The UK is spending ยฃ23M on AI tutors for 450,000 pupils because it can't hire enough teachers. It's the first major government positioning AI as a workforce replacement in schools.
While most schools have vague AI guidelines, Yellow Springs just approved a policy with actual teeth โ plagiarism enforcement, data bans, and a community workgroup. Here's why it matters and what NZ can learn.
Free Gemini Pro for a year. No card, no trial, no fine print. Every student 18+ can claim $200 of Google's best AI. Here's what's included and how to activate it.
MIT's new free program aims to make AI literacy universal โ no coding required.
An AI-native university where machines teach and humans supervise could reshape higher education globally.
UK schools are in a governance vacuum: no policy, widespread AI use, and teachers watching critical thinking erode in real time.
12 prompts that turn Claude into a subject-expert tutor for any grade level. Free. The $100/hour tutoring gap just closed.
While other countries patch together AI education piecemeal, South Korea is building it like infrastructure โ selecting 20 universities to create curricula for everyone.
Stanford finds AI reasoning errors cluster at early transition points then stay coherent but wrong. Benchmarks don't catch this. We're deploying systems that pass tests but fail reality.
Red means no AI. Yellow means cite it. Green means it's required. Here's how NZ schools can implement a stoplight framework for AI โ with templates you can use tomorrow.
While K-12 AI literacy dominates headlines, a law school is making AI education mandatory for all students โ setting a precedent for professional training everywhere.
While states race to mandate AI literacy, Pennsylvania's largest educators union is sounding the alarm about AI overreach in schools โ banning it from assessments and evaluations.
As schools worldwide grapple with AI, a new free course offers a practical middle path between banning and blindly adopting AI tools in education.
Purdue is the first US university to make AI competency a graduation requirement for every undergraduate โ engineers, artists, business majors, everyone. Fall 2026.
While most districts are still debating AI bans, Boston is putting AI literacy in every high school classroom starting September 2026 โ the first major US district to do it.
India's state boards give AI curriculum real teeth, Google scales to 400+ campuses, and the media literacy gap gets urgent.
Punjab is the first Indian state to tie AI performance directly to board certificate outcomes, giving its AI curriculum enforcement teeth that national policies lack.
A new framework from Student Defense argues students deserve rights around AI decision-making โ not just access to AI tools.
AI literacy isn't an elective anymore. Pepperdine is making it required for every undergraduate โ business majors, artists, biologists, everyone.
Over half of elementary teachers report students struggle to distinguish AI-generated from human-created content. The race to update curricula is on โ but AI isn't waiting.
While most states are still debating guidelines, Utah passed nine AI laws in a single session. Bell-to-bell phone bans, AI literacy mandates, deepfake protections, and health insurance guardrails โ all heading to the governor's desk.
The Khan TED Institute offers a $10K AI degree backed by Google and Microsoft, Utah builds an age-appropriate tech staircase, and Google scales to 400+ campuses.
Iceland is betting that teacher-first AI adoption at national scale can work where big-country budgets can't. Claude speaks Icelandic. The teachers get trained. The students come second โ and that's the point.
Easelearn's Immersive Classroom delivers real-time AI instruction with dynamic slides and interactive quizzes. Is this the future of education?
The students living through AI's takeover of classrooms are the ones sounding the loudest alarm โ and nobody seems to be listening.
The UK is pushing ahead with supervised AI tutors in schools while the U.S. pushes pause. Which approach will prove right?
No phones, limited AI for young kids, mandatory digital literacy by 7th grade, and a social media tax to pay for the damage. Utah isn't waiting for Washington.
Florida's largest school district draws a hard line on AI โ bans on tests, mandatory citation, and criminal deepfake penalties.
The era of AI-in-education guidelines is over. States are writing actual laws, and they don't agree on much.
Over 250 experts say generative AI shouldn't be anywhere near a classroom until independent safety audits are done. The call directly challenges the rush to integrate AI into education.
New Zealand has no AI-specific legislation, but that doesn't mean no obligations exist. Here's what businesses actually need to do in 2026.
NZ and Australia take different approaches to AI regulation. If you work across both countries, here's what you need to know.
The Privacy Commissioner has clear expectations for AI use. Most professionals are getting it wrong. Here's the checklist.
Boston is putting real money behind AI literacy for every student. Meanwhile, New Zealand schools still have no national AI education framework.
ASU President Michael Crow wants AI to empower individuals, not replace them. will.i.am wants every person to have an 'agentic self.' The summit talked equity โ but who's actually delivering it?
Minnesota wants to put AI surveillance in eight school districts to detect weapons before violence happens. The ACLU says the technology lacks evidence and could disproportionately harm vulnerable students.
NJ isn't teaching kids to use AI โ it's teaching them to survive it. The state's 'Age of Fakery' curriculum reframes AI literacy as information defense.
Red means banned. Yellow means proceed with caution. Green means go. NYC's traffic light framework is the clearest district AI policy in the US โ and it could become the template everyone copies.
Ohio didn't just publish guidelines โ it set a deadline. Every K-12 district in the state must have an AI policy in place by July 1, making Ohio the first U.S. state to turn AI education policy from optional into mandatory.
UK teachers are adopting AI faster than policy can keep up, and confidence is lagging far behind usage. The pattern is global โ but so are the solutions.
While the US and China push national AI curriculum mandates, Latin America is trying something else: a regional observatory to coordinate policy, share evidence, and keep inequalities from deepening.
The generation expected to embrace AI is pushing back. Gallup finds Gen Z increasingly skeptical, with 83% saying AI will make learning harder โ not easier.
For the first time, Japan's approved high school textbooks contain substantive AI content spanning 67 passages across subjects from civics to geography. It's the moment where AI education stops being a policy discussion and becomes a lesson plan.
A former YC founder runs her homeschool and household on 11 named AI agents. Full Montessori curriculum: $8. Grocery orders via voice note. Scheduled 'ignore the kids' time. It's either the future of parenting or its most dystopian twist.
OpenAI's new ChatGPT for Teachers puts FERPA-compliant AI in K-12 classrooms for free through 2027. Is this the breakthrough education has been waiting for โ or a Trojan horse?
South Korea just became the first country to classify AI student evaluation as high-risk โ mandatory human oversight, transparency, and impact assessments for any AI that grades students.
After a year of state-level patchwork, Washington has issued its first federal AI education priority. It's a grant preference โ not a mandate โ and the tension between the two approaches says everything about where AI policy is headed.
The first peer-reviewed study linking AI dependence in education to measurable social skill decline. 608 students surveyed. The correlation is real. But there's a mitigating factor.
Alpha Schools charges $55K/year for AI-driven education with no teachers. It's expanding to Chicago this fall. Researchers worry about equity, scalability, and what kids lose.
While Western schools argue about AI in classrooms, China is executing a coordinated national rollout โ mandatory AI literacy for all students, AI teacher certification, and AI-integrated campuses by 2030.
Students are acing exams with Google Lens โ and teachers can see it happening in real time. LA's AI cheating scandal is the canary in the coal mine for every school district that gave kids AI tools without changing how they test.
Schools are writing AI policies faster than they're training teachers. Most educators still have no professional development on AI, and the poorest schools are being left furthest behind.
Federal inaction has created a vacuum โ and 31 states are filling it with 134 AI education bills. Idaho just signed the first one. The patchwork has begun.
India's CBSE board is making AI and computational thinking mandatory for students aged 8-14 across 28,000 schools. The largest mandatory AI education program in the world starts next year.
Denver Public Schools has approved student access to Google Gemini, NotebookLM, and MagicSchool on school devices. Teachers say the rollout caught them off guard.
A $23M partnership is training 400,000 teachers to use agentic AI โ with teachers training teachers, not tech companies. The goal: keep educators in the driver's seat for AI in schools.
Boston Public Schools will become the first major US city school district to launch a comprehensive AI literacy curriculum. Mayor Michelle Wu announced the initiative at Eliot K-8 Innovation Upper...
The University of Texas at Arlington has launched a four-session AI webinar series designed to help educators integrate AI into their classrooms. The series covers practical, ethical, and creative...
India's Central Board of Secondary Education (CBSE) has introduced AI and computational thinking into the curriculum for Classes 3-8. The move makes India one of the first countries to mandate AI...
The first week of April 2026 saw major developments in AI education policy globally. India's CBSE mandate (Classes 3-8), Boston's AI literacy curriculum, and UTA's teacher training series represent a...
Four AI education stories you need to know.
Anthropic announced 'no surcharge' for 1M context in March 2026. Developers are being billed for 'extra usage' instead. The hidden costs of large context windows.
OpenAI's historic $122 billion funding round has implications for education: as AI companies secure unprecedented capital, questions emerge about who pays for AI literacy training.
An AI-written research paper passing peer review (reported March 30) has significant implications for education. If AI can write publishable scientific papers, what should students learn about...
UK trains 10 million workers, teachers move beyond AI basics, and the NSF funds $11M for K-12 AI education.
